plots.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "addeddb1",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-22T13:36:26.454251Z",
     "start_time": "2022-04-22T13:36:26.361572Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The autoreload extension is already loaded. To reload it, use:\n",
      "  %reload_ext autoreload\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import pickle\n",
    "from tqdm import tqdm\n",
    "from scipy.optimize import curve_fit\n",
    "from matplotlib import pyplot as plt\n",
    "from IPython.display import set_matplotlib_formats\n",
    "#set_matplotlib_formats('png', 'pdf')\n",
    "import numpy as np\n",
    "import matplotlib as mpl\n",
    "import gzip as gz\n",
    "import sys\n",
    "mpl.rcParams.update(mpl.rcParamsDefault)\n",
    "\n",
    "width = 3.487 * 1.5 * 1.1\n",
    "height = width / 1.618 * 1.2\n",
    "\n",
    "plt.rc('font', family='sans-serif')#, serif='Times')\n",
    "plt.rc('xtick', labelsize=10)\n",
    "plt.rc('ytick', labelsize=10)\n",
    "plt.rc('axes', labelsize=12)\n",
    "plt.rc('legend', fontsize=10)\n",
    "#plt.rcParams['mathtext.fontset'] = 'stix'\n",
    "#plt.rcParams['font.family'] = 'STIXGeneral'\n",
    "params = {\n",
    "'text.usetex': True,\n",
    "#'font.size': 15,\n",
    "#'text.latex.unicode': True\n",
    "}\n",
    "plt.rcParams.update(params)\n",
    "\n",
    "#fig, axes = plt.subplots(1, 1, figsize=(width, height))\n",
    "#ax = axes\n",
    "#ax.set_title('RefSeq assembly graph')\n",
    "\n",
    "# colors = ['c', 'b', 'g', 'r']\n",
    "\n",
    "%matplotlib inline\n",
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "1159b6aa",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-22T13:36:29.398161Z",
     "start_time": "2022-04-22T13:36:29.374391Z"
    }
   },
   "outputs": [],
   "source": [
    "def to_seconds(s):\n",
    "    parts = s.split(':')\n",
    "    assert len(parts) <= 3\n",
    "    time = 0.\n",
    "    for p in parts:\n",
    "        time = time * 60 + float(p)\n",
    "    return time"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aa49ed37",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-08T10:01:23.424239Z",
     "start_time": "2021-10-08T10:01:20.108060Z"
    }
   },
   "source": [
    "np.random.seed(10)\n",
    "with open('/cluster/home/mikhaika/metagenome/data/coordinates_K/data/random_1M.fasta', 'w') as f:\n",
    "    for i in range(1000000):\n",
    "        kmer = ''.join(np.random.choice(['A', 'C', 'G', 'T'], 21))\n",
    "        count = np.random.randint(1, 1000)\n",
    "        f.write(f'>\\n{kmer}\\n' * count)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "9f285dd6",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-22T13:36:30.200207Z",
     "start_time": "2022-04-22T13:36:30.177042Z"
    }
   },
   "outputs": [],
   "source": [
    "def get_size(fname):\n",
    "    try:\n",
    "        return os.path.getsize(fname)\n",
    "    except:\n",
    "        return np.nan"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a259aa99",
   "metadata": {},
   "source": [
    "## PacBio SMRT Viruses reads"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "aa0d8c90",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:17.657643Z",
     "start_time": "2022-04-01T22:36:17.546620Z"
    },
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "with open('gzip_spring_size.pkl', 'rb') as f:\n",
    "    gzip, spring_size, gzip_size = pickle.load(f)\n",
    "\n",
    "# gzip = !find ~/metagenome/data/hifi_sra/viruses_hifi_data/compressed -name \"*_no_header.fasta.gz\"\n",
    "\n",
    "# spring_size = []\n",
    "# gzip_size = []\n",
    "\n",
    "# for x in tqdm(gzip):\n",
    "#     gzip_size.append(get_size(x))\n",
    "#     ID = x[:-len('_no_header.fasta.gz')]\n",
    "#     spring_fname = f'{ID}_no_header.spring'\n",
    "#     spring_size.append(get_size(spring_fname))\n",
    "\n",
    "# spring_size = np.array(spring_size)\n",
    "# gzip_size = np.array(gzip_size)\n",
    "\n",
    "# with open('gzip_spring_size.pkl', 'wb') as f:\n",
    "#     pickle.dump((gzip, spring_size, gzip_size), f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "ae45466d",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:18.374915Z",
     "start_time": "2022-04-01T22:36:18.336702Z"
    }
   },
   "outputs": [],
   "source": [
    "with open('hifi_score_ratio.pkl', 'rb') as f:\n",
    "    score_ratio = pickle.load(f)\n",
    "\n",
    "# score_ratio = []\n",
    "\n",
    "# for x in tqdm(gzip, mininterval=5):\n",
    "#     ID = x[:-len('_no_header.fasta.gz')]\n",
    "#     SRA = ID.split('/')[-1]\n",
    "#     file = f'/cluster/home/mikhaika/metagenome/raw_data/hifi_sra/viruses_hifi_data/{SRA}/{SRA}.fastq.gz'\n",
    "#     scores = ! zless $file | paste - - - - | cut -f4 | head -n 100\n",
    "#     score_ratio.append(np.mean([s == '~' for s in ''.join(scores)]))\n",
    "\n",
    "# score_ratio = np.array(score_ratio)\n",
    "# with open('hifi_score_ratio.pkl', 'wb') as f:\n",
    "#     pickle.dump(score_ratio, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "c82fff4b",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:18.922091Z",
     "start_time": "2022-04-01T22:36:18.883641Z"
    }
   },
   "outputs": [],
   "source": [
    "with open('num_bp.pkl', 'rb') as f:\n",
    "    num_bp = pickle.load(f)\n",
    "\n",
    "# num_bp = []\n",
    "\n",
    "# for x in tqdm(gzip, mininterval=5):\n",
    "#     fname = x[:-len('_no_header.fasta.gz')] + '_num_bp'\n",
    "#     with open(fname, 'r') as f:\n",
    "#         n = int(f.readline())\n",
    "#         num_bp.append(n)\n",
    "\n",
    "# num_bp = np.array(num_bp)\n",
    "\n",
    "# with open('num_bp.pkl', 'wb') as f:\n",
    "#     pickle.dump(num_bp, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "f5cd92dc",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:19.450515Z",
     "start_time": "2022-04-01T22:36:19.414664Z"
    }
   },
   "outputs": [],
   "source": [
    "with open('mtg_size.pkl', 'rb') as f:\n",
    "    mtg_size = pickle.load(f)\n",
    "\n",
    "# mtg_size = []\n",
    "\n",
    "# DIR = !echo $HOME\n",
    "# DIR = f'{DIR[0]}/metagenome/data/hifi_sra/viruses_hifi_data/mtg'\n",
    "\n",
    "# for x in gzip:\n",
    "#     ID = x.split('/')[-1][:-len('_no_header.fasta.gz')]\n",
    "#     graph_size = get_size(f'{DIR}/{ID[:6]}/{ID}/graph_small.dbg')\n",
    "#     anno_size = get_size(f'{DIR}/{ID[:6]}/{ID}/annotation.row_diff_coord.annodbg')\n",
    "#     mtg_size.append(graph_size + anno_size)\n",
    "\n",
    "# mtg_size = np.array(mtg_size)\n",
    "\n",
    "# with open('mtg_size.pkl', 'wb') as f:\n",
    "#     pickle.dump(mtg_size, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "831d9053",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:20.161981Z",
     "start_time": "2022-04-01T22:36:20.125833Z"
    }
   },
   "outputs": [],
   "source": [
    "with open('mtg_size_graph.pkl', 'rb') as f:\n",
    "    mtg_size_graph = pickle.load(f)\n",
    "\n",
    "# mtg_size_graph = []\n",
    "\n",
    "# DIR = !echo $HOME\n",
    "# DIR = f'{DIR[0]}/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/mtg'\n",
    "\n",
    "# for x in gzip:\n",
    "#     ID = x.split('/')[-1][:-len('_no_header.fasta.gz')]\n",
    "#     mtg_size_graph.append(get_size(f'{DIR}/{ID[:6]}/{ID}/graph_small.dbg'))\n",
    "\n",
    "# mtg_size_graph = np.array(mtg_size_graph)\n",
    "\n",
    "# with open('mtg_size_graph.pkl', 'wb') as f:\n",
    "#     pickle.dump(mtg_size_graph, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "a1a2acbd",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:20.778732Z",
     "start_time": "2022-04-01T22:36:20.741770Z"
    }
   },
   "outputs": [],
   "source": [
    "with open('mtg_size_anno.pkl', 'rb') as f:\n",
    "    mtg_size_anno = pickle.load(f)\n",
    "\n",
    "# mtg_size_anno = []\n",
    "\n",
    "# DIR = !echo $HOME\n",
    "# DIR = f'{DIR[0]}/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/mtg'\n",
    "\n",
    "# for x in gzip:\n",
    "#     ID = x.split('/')[-1][:-len('_no_header.fasta.gz')]\n",
    "#     mtg_size_anno.append(get_size(f'{DIR}/{ID[:6]}/{ID}/annotation.row_diff_coord.annodbg'))\n",
    "\n",
    "# mtg_size_anno = np.array(mtg_size_anno)\n",
    "\n",
    "# with open('mtg_size_anno.pkl', 'wb') as f:\n",
    "#     pickle.dump(mtg_size_anno, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "a90f1ce3",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:21.526690Z",
     "start_time": "2022-04-01T22:36:21.491789Z"
    }
   },
   "outputs": [],
   "source": [
    "with open('mtg2_size.pkl', 'rb') as f:\n",
    "    mtg2_size = pickle.load(f)\n",
    "\n",
    "# mtg2_size = []\n",
    "\n",
    "# DIR = !echo $HOME\n",
    "# DIR = f'{DIR[0]}/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/mtg_fork_opt'\n",
    "\n",
    "# for x in gzip:\n",
    "#     ID = x.split('/')[-1][:-len('_no_header.fasta.gz')]\n",
    "#     graph_size = get_size(f'{DIR}/{ID[:6]}/{ID}/graph_small.dbg')\n",
    "#     anno_size = get_size(f'{DIR}/{ID[:6]}/{ID}/annotation.row_diff_coord.annodbg')\n",
    "#     mtg2_size.append(graph_size + anno_size)\n",
    "\n",
    "# mtg2_size = np.array(mtg2_size)\n",
    "\n",
    "# with open('mtg2_size.pkl', 'wb') as f:\n",
    "#     pickle.dump(mtg2_size, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "48eee7cf",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:22.121466Z",
     "start_time": "2022-04-01T22:36:22.083854Z"
    }
   },
   "outputs": [],
   "source": [
    "with open('pfish_size.pkl', 'rb') as f:\n",
    "    pfish_size = pickle.load(f)\n",
    "\n",
    "# from pathlib import Path\n",
    "\n",
    "# pfish_size = []\n",
    "\n",
    "# DIR = !echo $HOME\n",
    "# DIR = f'{DIR[0]}/metagenome/data/hifi_sra/viruses_hifi_data/pufferfish_sparse'\n",
    "\n",
    "# for x in tqdm(gzip, mininterval=5):\n",
    "#     ID = x.split('/')[-1][:-len('_no_header.fasta.gz')]\n",
    "#     pfish_size.append(sum(f.stat().st_size for f in Path(f'{DIR}/{ID[:6]}/{ID}/{ID}').glob('*') if f.is_file()))\n",
    "\n",
    "# pfish_size = np.array(pfish_size)\n",
    "\n",
    "# with open('pfish_size.pkl', 'wb') as f:\n",
    "#     pickle.dump(pfish_size, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "5238f351",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:22.611761Z",
     "start_time": "2022-04-01T22:36:22.572974Z"
    }
   },
   "outputs": [],
   "source": [
    "with open('blast_size.pkl', 'rb') as f:\n",
    "    blast_size = pickle.load(f)\n",
    "\n",
    "# from pathlib import Path\n",
    "\n",
    "# blast_size = []\n",
    "\n",
    "# DIR = !echo $HOME\n",
    "# DIR = f'{DIR[0]}/metagenome/data/hifi_sra/viruses_hifi_data/blast'\n",
    "\n",
    "# for x in tqdm(gzip):\n",
    "#     ID = x.split('/')[-1][:-len('_no_header.fasta.gz')]\n",
    "#     blast_size.append(sum(f.stat().st_size for f in Path(f'{DIR}/{ID[:6]}/{ID}').glob('*') if f.is_file()))\n",
    "\n",
    "# blast_size = np.array(blast_size)\n",
    "\n",
    "# with open('blast_size.pkl', 'wb') as f:\n",
    "#     pickle.dump(blast_size, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "9261d1df",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:23.855207Z",
     "start_time": "2022-04-01T22:36:23.816565Z"
    }
   },
   "outputs": [],
   "source": [
    "with open('mgblst_size.pkl', 'rb') as f:\n",
    "    mgblst_size = pickle.load(f)\n",
    "\n",
    "# from pathlib import Path\n",
    "\n",
    "# mgblst_size = []\n",
    "\n",
    "# DIR = !echo $HOME\n",
    "# DIR = f'{DIR[0]}/metagenome/data/hifi_sra/viruses_hifi_data/megablast'\n",
    "\n",
    "# for x in tqdm(gzip, mininterval=5):\n",
    "#     ID = x.split('/')[-1][:-len('_no_header.fasta.gz')]\n",
    "#     mgblst_size.append(sum(f.stat().st_size for f in Path(f'{DIR}/{ID[:6]}/{ID}').glob('*') if f.is_file()))\n",
    "\n",
    "# mgblst_size = np.array(mgblst_size) + blast_size\n",
    "\n",
    "# with open('mgblst_size.pkl', 'wb') as f:\n",
    "#     pickle.dump(mgblst_size, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "07b89508",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:24.483454Z",
     "start_time": "2022-04-01T22:36:24.426837Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "149.967152078"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(gzip_size) / 1e9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "5c1470a5",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:25.084414Z",
     "start_time": "2022-04-01T22:36:25.031836Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "875.484502657"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(num_bp) / 1e9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "22ce3594",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:29.136227Z",
     "start_time": "2022-04-01T22:36:28.682322Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MegaBLAST compression: 0.1\n",
      "Pufferfish compression: 0.3\n",
      "BLAST compression: 3.1\n",
      "gzip -9 compression: 5.8\n",
      "mtg (single succ.) compression: 6.2\n",
      "mtg (multiple succ.) compression: 7.1\n",
      "Spring compression: 18.0\n"
     ]
    }
   ],
   "source": [
    "print(f\"MegaBLAST compression: {sum(num_bp) / sum(mgblst_size - blast_size):.1f}\")\n",
    "print(f\"Pufferfish compression: {sum(num_bp) / sum(pfish_size):.1f}\")\n",
    "print(f\"BLAST compression: {sum(num_bp) / sum(blast_size):.1f}\")\n",
    "print(f\"gzip -9 compression: {sum(num_bp) / sum(gzip_size):.1f}\")\n",
    "print(f\"mtg (single succ.) compression: {sum(num_bp) / sum(mtg_size):.1f}\")\n",
    "print(f\"mtg (multiple succ.) compression: {sum(num_bp[mtg2_size > 0]) / sum(mtg2_size[mtg2_size > 0]):.1f}\")\n",
    "print(f\"Spring compression: {sum(num_bp) / sum(spring_size):.1f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "c129057a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:30.071763Z",
     "start_time": "2022-04-01T22:36:29.713198Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MegaBLAST bits/bp: 105.352\n",
      "Pufferfish bits/bp: 25.113\n",
      "BLAST bits/bp: 2.580\n",
      "gzip -9 bits/bp: 1.370\n",
      "mtg (single succ.) bits/bp: 1.300\n",
      "mtg (multiple succ.) bits/bp: 1.130\n",
      "Spring bits/bp: 0.445\n"
     ]
    }
   ],
   "source": [
    "print(f\"MegaBLAST bits/bp: {sum(mgblst_size - blast_size) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"Pufferfish bits/bp: {sum(pfish_size) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"BLAST bits/bp: {sum(blast_size) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"gzip -9 bits/bp: {sum(gzip_size) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"mtg (single succ.) bits/bp: {sum(mtg_size) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"mtg (multiple succ.) bits/bp: {sum(mtg2_size[mtg2_size > 0]) * 8 / sum(num_bp[mtg2_size > 0]):.3f}\")\n",
    "print(f\"Spring bits/bp: {sum(spring_size) * 8 / sum(num_bp):.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "6807effa",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:31.610803Z",
     "start_time": "2022-04-01T22:36:31.582740Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "152884"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(score_ratio)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cee58ede",
   "metadata": {},
   "source": [
    "### HiFi reads"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c26f608c",
   "metadata": {},
   "source": [
    "Now out of all 152,884 PacBio SMRT samples, we are going to select only the Hi-Fi ones.\n",
    "This is not a property written in the metadata at NCBI SRA, so we're going to use the fact that for Hi-Fi sequencing, most of the sequencing scores are '\\~'. Thus, we're going to look at the frequency of '\\~' quality scores in each sample and filter out those samples for which this frequency is less than 0.7."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "03641724",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:38.926734Z",
     "start_time": "2022-04-01T22:36:35.666505Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 376.596x232.754 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "width = 3.487 * 1.5 * 1\n",
    "height = width / 1.618\n",
    "\n",
    "plt.figure(figsize=(width, height))\n",
    "plt.hist(score_ratio, bins=100)\n",
    "plt.xlabel('fraction of tilda scores')\n",
    "plt.ylabel('\\# read sets')\n",
    "plt.grid()\n",
    "plt.tight_layout()\n",
    "plt.savefig('hifi_score_frequency.pdf', format='pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c3dd9c25",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-22T14:01:27.717591Z",
     "start_time": "2021-10-22T14:01:27.687609Z"
    }
   },
   "source": [
    "The fraction of libraries we keep:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "255041f2",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:38.951695Z",
     "start_time": "2022-04-01T22:36:38.929720Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "99.70%\n"
     ]
    }
   ],
   "source": [
    "print(f'{np.mean(score_ratio > 0.7) * 100:.2f}%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "ff17867c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:39.247516Z",
     "start_time": "2022-04-01T22:36:38.956001Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "152418"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(score_ratio > 0.7)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "c8d921fe",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:39.320067Z",
     "start_time": "2022-04-01T22:36:39.251631Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "38.4434618997873"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(num_bp[score_ratio > 0.7]) / sum(spring_size[score_ratio > 0.7])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0619a934",
   "metadata": {},
   "source": [
    "As we can see, only 0.3% of samples were discarded by this criterion:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "822619af",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:39.457479Z",
     "start_time": "2022-04-01T22:36:39.326006Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[['SRR3474205' '0.0']\n",
      " ['SRR3474206' '0.0']\n",
      " ['SRR3474203' '0.0']\n",
      " ['SRR3474204' '0.0']\n",
      " ['SRR2041922' '0.0']\n",
      " ['SRR2042468' '0.0']\n",
      " ['SRR7892517' '0.0']\n",
      " ['SRR528818' '0.0']\n",
      " ['SRR528803' '0.012313836425659372']\n",
      " ['SRR528788' '4.7216582463761274e-05']\n",
      " ['SRR528812' '0.000330531683822835']\n",
      " ['SRR528819' '8.756567425569177e-05']\n",
      " ['SRR528843' '0.0']\n",
      " ['SRR528777' '0.004063148093283109']\n",
      " ['SRR528808' '0.0']\n",
      " ['SRR528796' '0.0']\n",
      " ['SRR528798' '0.0003410932037179159']\n",
      " ['SRR528838' '0.0004110903028365231']\n",
      " ['SRR528842' '0.0018848700967906266']\n",
      " ['SRR528846' '4.169967891247237e-05']\n",
      " ['SRR528831' '0.0']\n",
      " ['SRR528826' '0.0']\n",
      " ['SRR528837' '0.0']\n",
      " ['SRR528833' '0.0']\n",
      " ['SRR528850' '4.5724737082761774e-05']\n",
      " ['SRR528851' '0.004765943697241407']\n",
      " ['SRR528834' '0.0']\n",
      " ['SRR528840' '0.005685971895053204']\n",
      " ['SRR528802' '0.006106529975484734']\n",
      " ['SRR528828' '0.013454861111111112']\n",
      " ['SRR528852' '0.003726272412432598']\n",
      " ['SRR528835' '0.00021592675764380722']\n",
      " ['SRR528773' '0.0']\n",
      " ['SRR528774' '0.0002110261144816671']\n",
      " ['SRR528821' '0.0']\n",
      " ['SRR528797' '0.0']\n",
      " ['SRR528779' '0.005623003194888179']\n",
      " ['SRR528799' '0.0006172023101000749']\n",
      " ['SRR528783' '0.0']\n",
      " ['SRR528830' '0.002552204176334107']\n",
      " ['SRR528832' '0.0']\n",
      " ['SRR528810' '9.33009889904833e-05']\n",
      " ['SRR528805' '0.0']\n",
      " ['SRR528781' '0.0037033883875361826']\n",
      " ['SRR528829' '0.003840142889037732']\n",
      " ['SRR528816' '0.00711699338244475']\n",
      " ['SRR528790' '4.348771472059143e-05']\n",
      " ['SRR528791' '0.0']\n",
      " ['SRR528794' '0.000434824620736303']\n",
      " ['SRR528786' '0.0']\n",
      " ['SRR528822' '0.0']\n",
      " ['SRR528813' '0.005290748471022231']\n",
      " ['SRR528825' '0.0']\n",
      " ['SRR528782' '0.00013240940989539657']\n",
      " ['SRR528784' '0.0']\n",
      " ['SRR528807' '0.0']\n",
      " ['SRR528785' '0.0']\n",
      " ['SRR528792' '0.01160426481527399']\n",
      " ['SRR528789' '0.0']\n",
      " ['SRR528800' '0.005330388013830196']\n",
      " ['SRR528811' '0.0']\n",
      " ['SRR528844' '0.0']\n",
      " ['SRR528778' '0.009758497683381852']\n",
      " ['SRR528787' '0.0']\n",
      " ['SRR528847' '0.0']\n",
      " ['SRR528814' '0.010677360131413663']\n",
      " ['SRR528827' '0.009347753185781354']\n",
      " ['SRR528815' '0.00014349947383526261']\n",
      " ['SRR528823' '0.0']\n",
      " ['SRR528795' '0.020664281034222066']\n",
      " ['SRR528817' '0.0']\n",
      " ['SRR528776' '0.0026750456456201433']\n",
      " ['SRR528839' '0.004344646621069385']\n",
      " ['SRR528841' '0.00039209920109787774']\n",
      " ['SRR528801' '0.007307656953764133']\n",
      " ['SRR528836' '0.0']\n",
      " ['SRR528848' '0.0']\n",
      " ['SRR528845' '4.0498947027377285e-05']\n",
      " ['SRR528809' '0.0013437364542696142']\n",
      " ['SRR528780' '0.001470129638704504']\n",
      " ['SRR528820' '0.0']\n",
      " ['SRR528824' '0.0']\n",
      " ['SRR528849' '0.0']\n",
      " ['SRR528775' '0.004971286534670438']\n",
      " ['SRR528806' '0.0005354274495805819']\n",
      " ['SRR528772' '0.001183431952662722']\n",
      " ['SRR528804' '0.00014357501794687725']\n",
      " ['SRR528793' '0.006513894746707906']\n",
      " ['ERR1223793' '0.0']\n",
      " ['ERR1223795' '0.0']\n",
      " ['ERR1223796' '0.0']\n",
      " ['ERR1223794' '0.0']\n",
      " ['SRR530880' '0.0']\n",
      " ['SRR530881' '0.0']\n",
      " ['SRR530879' '0.0']\n",
      " ['SRR13587452' '0.5030821497994852']\n",
      " ['SRR13587445' '0.20904349280611095']\n",
      " ['SRR13587449' '0.15409143839716172']\n",
      " ['SRR13587443' '0.1337693625445785']\n",
      " ['SRR13587446' '0.10435234622724876']\n",
      " ['SRR13587450' '0.4296596666487638']\n",
      " ['SRR13587447' '0.1730288749430173']\n",
      " ['SRR13587448' '0.2673204069581716']\n",
      " ['SRR13587451' '0.3925533243986098']\n",
      " ['SRR13587444' '0.14208893008688042']\n",
      " ['SRR5122814' '0.0']\n",
      " ['SRR5122783' '0.0']\n",
      " ['SRR5122802' '0.0']\n",
      " ['SRR5122817' '0.0']\n",
      " ['SRR5122793' '0.0']\n",
      " ['SRR5122799' '0.0']\n",
      " ['SRR5122786' '0.0']\n",
      " ['SRR5122796' '0.0']\n",
      " ['SRR5122831' '0.0']\n",
      " ['SRR5122789' '0.0']\n",
      " ['SRR5122797' '0.0']\n",
      " ['SRR5122800' '0.0']\n",
      " ['SRR5122782' '0.0']\n",
      " ['SRR5122790' '0.0']\n",
      " ['SRR5122816' '0.0']\n",
      " ['SRR5122828' '0.0']\n",
      " ['SRR5122812' '0.0']\n",
      " ['SRR5122830' '0.0']\n",
      " ['SRR5122807' '0.0']\n",
      " ['SRR5122806' '0.0']\n",
      " ['SRR5122785' '0.0']\n",
      " ['SRR5122795' '0.0']\n",
      " ['SRR5122787' '0.0']\n",
      " ['SRR5122820' '0.0']\n",
      " ['SRR5122803' '0.0']\n",
      " ['SRR5122829' '0.0']\n",
      " ['SRR5122805' '0.0']\n",
      " ['SRR5122826' '0.0']\n",
      " ['SRR5122825' '0.0']\n",
      " ['SRR5122827' '0.0']\n",
      " ['SRR5122815' '0.0']\n",
      " ['SRR5122798' '0.0']\n",
      " ['SRR5122824' '0.0']\n",
      " ['SRR5122821' '0.0']\n",
      " ['SRR5122811' '0.0']\n",
      " ['SRR5122801' '0.0']\n",
      " ['SRR5122822' '0.0']\n",
      " ['SRR5122788' '0.0']\n",
      " ['SRR5122792' '0.0']\n",
      " ['SRR5122810' '0.0']\n",
      " ['SRR5122784' '0.0']\n",
      " ['SRR5122794' '0.0']\n",
      " ['SRR5122818' '0.0']\n",
      " ['SRR5122823' '0.0']\n",
      " ['SRR5122804' '0.0']\n",
      " ['SRR5122808' '0.0']\n",
      " ['SRR5122813' '0.0']\n",
      " ['SRR5122809' '0.0']\n",
      " ['SRR5122819' '0.0']\n",
      " ['SRR5122791' '0.0']\n",
      " ['SRR1640361' '0.0']\n",
      " ['SRR1640364' '0.0']\n",
      " ['SRR1640362' '0.0']\n",
      " ['SRR1640363' '0.0']\n",
      " ['SRR15157490' '0.6876163384530207']\n",
      " ['SRR16031424' '0.6774074646564332']\n",
      " ['ERR2270747' '0.6277316441449566']\n",
      " ['ERR2643772' '0.0']\n",
      " ['ERR2643773' '0.5088229041173552']\n",
      " ['ERR2643771' '0.0']\n",
      " ['ERR2106423' '0.0']\n",
      " ['ERR2106426' '0.0']\n",
      " ['ERR2106421' '0.0']\n",
      " ['ERR2106424' '0.0']\n",
      " ['ERR2106427' '0.0']\n",
      " ['ERR2106428' '0.0']\n",
      " ['ERR2106425' '0.0']\n",
      " ['ERR2106422' '0.0']\n",
      " ['SRR6917576' '0.0']\n",
      " ['SRR13736174' '0.6761586752642742']\n",
      " ['ERR3278853' '0.0']\n",
      " ['DRR014572' '0.0']\n",
      " ['DRR014571' '0.0']\n",
      " ['SRR1801372' '0.0']\n",
      " ['SRR1801364' '0.0']\n",
      " ['SRR1801365' '0.0']\n",
      " ['SRR1801367' '0.0']\n",
      " ['SRR1801368' '0.0']\n",
      " ['SRR1801369' '0.0']\n",
      " ['SRR1801366' '0.0']\n",
      " ['SRR5429726' '0.0']\n",
      " ['SRR5429728' '0.0']\n",
      " ['SRR13380676' '0.0']\n",
      " ['SRR13380672' '0.0']\n",
      " ['SRR13380671' '0.0']\n",
      " ['SRR13380670' '0.0']\n",
      " ['SRR13380667' '0.0']\n",
      " ['SRR13380665' '0.0']\n",
      " ['SRR13380669' '0.0']\n",
      " ['SRR13380675' '0.0']\n",
      " ['SRR13380673' '0.0']\n",
      " ['SRR13380666' '0.0']\n",
      " ['SRR13380674' '0.0']\n",
      " ['SRR5559126' '0.0']\n",
      " ['SRR5559127' '0.0']\n",
      " ['SRR11171704' '0.2762613592942701']\n",
      " ['SRR11171707' '0.27485813066713627']\n",
      " ['SRR11171706' '0.48870420227086503']\n",
      " ['SRR11171703' '0.3627130641420238']\n",
      " ['SRR11171711' '0.0']\n",
      " ['SRR11171709' '0.2594387230164718']\n",
      " ['SRR11171710' '0.24382310557146333']\n",
      " ['SRR11171705' '0.4038882122605063']\n",
      " ['SRR11171708' '0.22884325279702294']\n",
      " ['SRR2126279' '0.0']\n",
      " ['SRR2126028' '0.0']\n",
      " ['SRR2126272' '0.0']\n",
      " ['SRR2126271' '0.0']\n",
      " ['SRR2126274' '0.0']\n",
      " ['SRR2126027' '0.0']\n",
      " ['SRR2126276' '0.0']\n",
      " ['SRR2126277' '0.0']\n",
      " ['SRR2126275' '0.0']\n",
      " ['SRR2126003' '0.0']\n",
      " ['SRR2126273' '0.0']\n",
      " ['SRR2126278' '0.0']\n",
      " ['SRR3140624' '0.0']\n",
      " ['SRR3140630' '0.0']\n",
      " ['SRR3140626' '0.0']\n",
      " ['SRR3141603' '0.0']\n",
      " ['SRR3140628' '0.0']\n",
      " ['SRR3140629' '0.0']\n",
      " ['SRR955463' '0.0']\n",
      " ['SRR955464' '0.0']\n",
      " ['ERR2528475' '0.0']\n",
      " ['SRR15935422' '0.6833669837437296']\n",
      " ['SRR15935421' '0.6773531183266714']\n",
      " ['SRR15938969' '0.5566083770283818']\n",
      " ['ERR2503173' '0.0']\n",
      " ['ERR2379737' '0.4190828003707185']\n",
      " ['SRR13144523' '0.6985455986508686']\n",
      " ['SRR13125403' '0.5006394344306204']\n",
      " ['SRR13125402' '0.49168299157287093']\n",
      " ['SRR13125398' '0.5298677582125936']\n",
      " ['SRR13125397' '0.49831815797045037']\n",
      " ['SRR13125400' '0.3932326796165944']\n",
      " ['SRR6049445' '0.0']\n",
      " ['SRR6049447' '0.0']\n",
      " ['SRR6049443' '0.0']\n",
      " ['SRR6049441' '0.0']\n",
      " ['SRR6049435' '0.0']\n",
      " ['SRR6049438' '0.0']\n",
      " ['SRR15796789' '0.6867259793528444']\n",
      " ['SRR513761' '0.0']\n",
      " ['SRR513763' '0.0']\n",
      " ['SRR513762' '0.0']\n",
      " ['SRR12218891' '0.0']\n",
      " ['SRR12218892' '0.08845884616641887']\n",
      " ['SRR12218884' '0.0']\n",
      " ['SRR12218882' '0.0']\n",
      " ['SRR12218893' '0.0']\n",
      " ['SRR12218894' '0.0']\n",
      " ['SRR12218889' '0.0']\n",
      " ['SRR12218876' '0.0']\n",
      " ['SRR12218877' '0.0']\n",
      " ['SRR12218887' '0.0']\n",
      " ['SRR12218886' '0.0']\n",
      " ['SRR12218871' '0.0']\n",
      " ['SRR12218885' '0.0']\n",
      " ['SRR12218883' '0.0']\n",
      " ['SRR12218873' '0.0']\n",
      " ['SRR12218888' '0.0']\n",
      " ['SRR12218890' '0.0']\n",
      " ['SRR9719184' '0.59059749519946']\n",
      " ['SRR9719185' '0.39674439518720933']\n",
      " ['SRR5569826' '0.0']\n",
      " ['SRR5569827' '0.0']\n",
      " ['SRR9192557' '0.0']\n",
      " ['SRR9192559' '0.0']\n",
      " ['SRR9192558' '0.0']\n",
      " ['SRR5448265' '0.0']\n",
      " ['SRR5448262' '0.0']\n",
      " ['SRR5448270' '0.0']\n",
      " ['SRR5448273' '0.0']\n",
      " ['SRR5448266' '0.0']\n",
      " ['SRR5448277' '0.0']\n",
      " ['SRR5448261' '0.0']\n",
      " ['SRR5448268' '0.0']\n",
      " ['SRR5448275' '0.0']\n",
      " ['SRR5448267' '0.0']\n",
      " ['SRR5448263' '0.0']\n",
      " ['SRR5448269' '0.0']\n",
      " ['SRR5448276' '0.0']\n",
      " ['SRR5448264' '0.0']\n",
      " ['SRR5448274' '0.0']\n",
      " ['SRR5448271' '0.0']\n",
      " ['SRR5448272' '0.0']\n",
      " ['SRR11426430' '0.680272302069723']\n",
      " ['SRR3458180' '0.0']\n",
      " ['SRR3458170' '0.0']\n",
      " ['SRR3458162' '0.0']\n",
      " ['SRR3458196' '0.0']\n",
      " ['SRR3458173' '0.0']\n",
      " ['SRR3458168' '0.0']\n",
      " ['SRR3458184' '0.0']\n",
      " ['SRR3458248' '0.0']\n",
      " ['SRR3458250' '0.0']\n",
      " ['SRR3458172' '0.0']\n",
      " ['SRR3458182' '0.0']\n",
      " ['SRR3458187' '0.0']\n",
      " ['SRR3458185' '0.0']\n",
      " ['SRR3458175' '0.0']\n",
      " ['SRR3458224' '0.0']\n",
      " ['SRR3458166' '0.0']\n",
      " ['SRR3458167' '0.0']\n",
      " ['SRR3458181' '0.0']\n",
      " ['SRR3458177' '0.0']\n",
      " ['SRR3458249' '0.0']\n",
      " ['SRR3458183' '0.0']\n",
      " ['SRR3458179' '0.0']\n",
      " ['SRR3458238' '0.0']\n",
      " ['SRR3458174' '0.0']\n",
      " ['SRR3458171' '0.0']\n",
      " ['SRR3458163' '0.0']\n",
      " ['SRR3458189' '0.0']\n",
      " ['SRR3458186' '0.0']\n",
      " ['SRR3458188' '0.0']\n",
      " ['SRR3458164' '0.0']\n",
      " ['SRR3458178' '0.0']\n",
      " ['SRR3458165' '0.0']\n",
      " ['SRR3458176' '0.0']\n",
      " ['SRR3458169' '0.0']\n",
      " ['DRR288166' '0.0']\n",
      " ['SRR504652' '0.0']\n",
      " ['SRR5383881' '0.0']\n",
      " ['SRR5383891' '0.0']\n",
      " ['SRR5383883' '0.0']\n",
      " ['SRR5383885' '0.0']\n",
      " ['SRR5383888' '0.0']\n",
      " ['SRR5383880' '0.0']\n",
      " ['SRR5383890' '0.0']\n",
      " ['SRR5383879' '0.0']\n",
      " ['SRR5383887' '0.0']\n",
      " ['SRR5383884' '0.0']\n",
      " ['SRR5383889' '0.0']\n",
      " ['SRR5383886' '0.0']\n",
      " ['SRR5383882' '0.0']\n",
      " ['SRR8280928' '0.6770915993713921']\n",
      " ['SRR8280929' '0.6623067403938031']\n",
      " ['DRR072352' '0.0']\n",
      " ['DRR072343' '0.0']\n",
      " ['DRR072348' '0.0']\n",
      " ['DRR072342' '0.0']\n",
      " ['DRR072350' '0.0']\n",
      " ['DRR072347' '0.0']\n",
      " ['DRR072351' '0.0']\n",
      " ['DRR072344' '0.0']\n",
      " ['DRR072346' '0.0']\n",
      " ['DRR072345' '0.0']\n",
      " ['DRR072349' '0.0']\n",
      " ['DRR072353' '0.0']\n",
      " ['SRR12424739' '0.0']\n",
      " ['ERR2535383' '0.6905825769916997']\n",
      " ['ERR2538120' '0.0']\n",
      " ['ERR2538124' '0.0']\n",
      " ['ERR2535382' '0.5905991387551096']\n",
      " ['ERR2538122' '0.0']\n",
      " ['ERR2538119' '0.0']\n",
      " ['ERR2535379' '0.630230933683297']\n",
      " ['ERR2535380' '0.577899425161262']\n",
      " ['ERR2535381' '0.5359427366357774']\n",
      " ['ERR2538121' '0.0']\n",
      " ['ERR2538125' '0.0']\n",
      " ['ERR2538123' '0.0']\n",
      " ['ERR2535377' '0.6009681479330041']\n",
      " ['ERR2581430' '0.0']\n",
      " ['ERR2581424' '0.0']\n",
      " ['ERR2581425' '0.0']\n",
      " ['ERR2581427' '0.0']\n",
      " ['ERR2581429' '0.0']\n",
      " ['ERR2581426' '0.0']\n",
      " ['ERR2581428' '0.0']\n",
      " ['ERR2581423' '0.0']\n",
      " ['SRR10566658' '0.0']\n",
      " ['SRR1796797' '0.0']\n",
      " ['SRR1796349' '0.0']\n",
      " ['SRR1796784' '0.0']\n",
      " ['SRR1796334' '0.0']\n",
      " ['SRR1796823' '0.0']\n",
      " ['SRR1796313' '0.0']\n",
      " ['SRR1796822' '0.0']\n",
      " ['SRR1796317' '0.0']\n",
      " ['SRR1796800' '0.0']\n",
      " ['SRR1796821' '0.0']\n",
      " ['SRR1796824' '0.0']\n",
      " ['SRR1796328' '0.0']\n",
      " ['SRR1796787' '0.0']\n",
      " ['SRR1796804' '0.0']\n",
      " ['SRR1796325' '0.0']\n",
      " ['SRR1796348' '0.0']\n",
      " ['SRR1796315' '0.0']\n",
      " ['SRR1796330' '0.0']\n",
      " ['SRR1796351' '0.0']\n",
      " ['SRR1796794' '0.0']\n",
      " ['SRR1796065' '0.0']\n",
      " ['SRR1796795' '0.0']\n",
      " ['SRR1796788' '0.0']\n",
      " ['SRR1796314' '0.0']\n",
      " ['SRR1796812' '0.0']\n",
      " ['SRR1796791' '0.0']\n",
      " ['SRR1796813' '0.0']\n",
      " ['SRR1796819' '0.0']\n",
      " ['SRR961669' '0.0']\n",
      " ['SRR3539911' '0.0']\n",
      " ['SRR3539916' '0.0']\n",
      " ['SRR3539913' '0.0']\n",
      " ['SRR3539910' '0.0']\n",
      " ['SRR3539912' '0.0']\n",
      " ['SRR3539909' '0.0']\n",
      " ['SRR6438485' '0.0']\n",
      " ['SRR6438488' '0.0']\n",
      " ['SRR6438482' '0.0']\n",
      " ['SRR6438487' '0.0']\n",
      " ['SRR6438483' '0.0']\n",
      " ['SRR6438491' '0.0']\n",
      " ['SRR6438489' '0.0']\n",
      " ['SRR6438484' '0.0']\n",
      " ['SRR6438490' '0.0']\n",
      " ['SRR6438486' '0.0']\n",
      " ['ERR2112226' '0.0']\n",
      " ['ERR2112233' '0.0']\n",
      " ['ERR2112228' '0.0']\n",
      " ['ERR2112230' '0.0']\n",
      " ['ERR2112227' '0.0']\n",
      " ['ERR2112229' '0.0']\n",
      " ['ERR2112231' '0.0']\n",
      " ['ERR2112232' '0.0']\n",
      " ['SRR5105967' '0.0']\n",
      " ['SRR5105957' '0.0']\n",
      " ['SRR5105958' '0.0']\n",
      " ['SRR5105965' '0.0']\n",
      " ['SRR5105964' '0.0']\n",
      " ['SRR5105962' '0.0']\n",
      " ['SRR5105960' '0.0']\n",
      " ['SRR5105966' '0.0']\n",
      " ['SRR11308343' '0.0']\n",
      " ['SRR11310657' '0.0']\n",
      " ['SRR11308346' '0.0']\n",
      " ['SRR11308348' '0.0']\n",
      " ['SRR11308351' '0.0']\n",
      " ['SRR11308342' '0.0']\n",
      " ['SRR11308341' '0.0']\n",
      " ['SRR11308345' '0.0']\n",
      " ['SRR11308349' '0.0']\n",
      " ['SRR11308344' '0.0']\n",
      " ['SRR11308347' '0.0']\n",
      " ['SRR11336162' '0.0']\n",
      " ['SRR11308350' '0.0']\n",
      " ['ERR5234718' '0.0']\n",
      " ['SRR1564119' '0.0']\n",
      " ['SRR1564129' '0.0']\n",
      " ['SRR1564125' '0.0']\n",
      " ['SRR1564130' '0.0']\n",
      " ['SRR1564122' '0.0']\n",
      " ['SRR1564128' '0.0']\n",
      " ['SRR1564120' '0.0']\n",
      " ['SRR1564126' '0.0']\n",
      " ['SRR1564124' '0.0']\n",
      " ['SRR1564127' '0.0']\n",
      " ['SRR1564123' '0.0']\n",
      " ['SRR1564121' '0.0']]\n"
     ]
    }
   ],
   "source": [
    "print(np.concatenate((np.array([x.split('/')[-1][:-len('_no_header.fasta.gz')] for x in gzip])[score_ratio <= 0.7].reshape(-1, 1), score_ratio[score_ratio <= 0.7].reshape(-1, 1)), axis=1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "5ed9f0c7",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:39.675712Z",
     "start_time": "2022-04-01T22:36:39.461357Z"
    }
   },
   "outputs": [],
   "source": [
    "idx = score_ratio > 0.7\n",
    "gzip = np.array(gzip)[idx]\n",
    "gzip_size = gzip_size[idx]\n",
    "num_bp = num_bp[idx]\n",
    "spring_size = spring_size[idx]\n",
    "pfish_size = pfish_size[idx]\n",
    "mtg_size = mtg_size[idx]\n",
    "mtg_size_graph = mtg_size_graph[idx]\n",
    "mtg_size_anno = mtg_size_anno[idx]\n",
    "mtg2_size = mtg2_size[idx]\n",
    "blast_size = blast_size[idx]\n",
    "mgblst_size = mgblst_size[idx]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "30526982",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:40.670070Z",
     "start_time": "2022-04-01T22:36:40.620378Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "111.599186766"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(gzip_size) / 1e9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "cc211f0c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:42.411850Z",
     "start_time": "2022-04-01T22:36:42.360430Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "716.726703751"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(num_bp) / 1e9"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "997fc48e",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:46.216869Z",
     "start_time": "2022-04-01T22:36:45.876951Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MegaBLAST compression: 0.1\n",
      "Pufferfish compression: 0.4\n",
      "BLAST compression: 3.0\n",
      "gzip -9 compression: 6.4\n",
      "mtg (single succ.) compression: 14.7\n",
      "mtg (multiple succ.) compression: nan\n",
      "Spring compression: 38.4\n"
     ]
    }
   ],
   "source": [
    "print(f\"MegaBLAST compression: {sum(num_bp) / sum(mgblst_size - blast_size):.1f}\")\n",
    "print(f\"Pufferfish compression: {sum(num_bp) / sum(pfish_size):.1f}\")\n",
    "print(f\"BLAST compression: {sum(num_bp) / sum(blast_size):.1f}\")\n",
    "print(f\"gzip -9 compression: {sum(num_bp) / sum(gzip_size):.1f}\")\n",
    "print(f\"mtg (single succ.) compression: {sum(num_bp) / sum(mtg_size):.1f}\")\n",
    "print(f\"mtg (multiple succ.) compression: {sum(num_bp) / sum(mtg2_size):.1f}\")\n",
    "print(f\"Spring compression: {sum(num_bp) / sum(spring_size):.1f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "880c0ce0",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:46.823554Z",
     "start_time": "2022-04-01T22:36:46.488602Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MegaBLAST bits/bp: 125.848\n",
      "Pufferfish bits/bp: 21.816\n",
      "BLAST bits/bp: 2.678\n",
      "gzip -9 bits/bp: 1.246\n",
      "mtg (single succ.) bits/bp: 0.544\n",
      "mtg (multiple succ.) bits/bp: nan\n",
      "Spring bits/bp: 0.208\n"
     ]
    }
   ],
   "source": [
    "print(f\"MegaBLAST bits/bp: {sum(mgblst_size - blast_size) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"Pufferfish bits/bp: {sum(pfish_size) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"BLAST bits/bp: {sum(blast_size) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"gzip -9 bits/bp: {sum(gzip_size) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"mtg (single succ.) bits/bp: {sum(mtg_size) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"mtg (multiple succ.) bits/bp: {sum(mtg2_size) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"Spring bits/bp: {sum(spring_size) * 8 / sum(num_bp):.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "fe96ab33",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-01T22:36:48.428809Z",
     "start_time": "2022-04-01T22:36:48.312386Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mtg (single succ.), graph only, bits/bp: 0.151\n",
      "mtg (single succ.), anno only, bits/bp: 0.393\n"
     ]
    }
   ],
   "source": [
    "print(f\"mtg (single succ.), graph only, bits/bp: {sum(mtg_size_graph) * 8 / sum(num_bp):.3f}\")\n",
    "print(f\"mtg (single succ.), anno only, bits/bp: {sum(mtg_size_anno) * 8 / sum(num_bp):.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "d114abad",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:29.196731Z",
     "start_time": "2021-11-09T14:14:24.508240Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 376.596x325.856 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "width = 3.487 * 1.5 * 1\n",
    "height = width / 1.618 * 1.4\n",
    "\n",
    "x = np.array(num_bp)\n",
    "y = np.array(mtg_size * 8 / num_bp)\n",
    "plt.figure(figsize=(width, height))\n",
    "plt.scatter(x, y, color='b', alpha=0.1)\n",
    "\n",
    "plt.title('PacBio HiFi (viruses)')\n",
    "plt.ylabel('Size, bits/bp')\n",
    "plt.xlabel('Library size, bp')\n",
    "plt.grid()\n",
    "plt.xscale('log')\n",
    "plt.tight_layout()\n",
    "plt.savefig('hifi_size_vs_numbp.pdf', format='pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "401e258a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:29.947676Z",
     "start_time": "2021-11-09T14:14:29.200851Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 376.596x325.856 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "width = 3.487 * 1.5 * 1\n",
    "height = width / 1.618 * 1.4\n",
    "\n",
    "x = np.array(num_bp)\n",
    "y = np.array(pfish_size * 8 / num_bp)\n",
    "plt.figure(figsize=(width, height))\n",
    "plt.scatter(x, y, color='b', alpha=0.1)\n",
    "\n",
    "plt.title('PacBio HiFi (viruses)')\n",
    "plt.ylabel('Pufferfish, bits/bp')\n",
    "plt.xlabel('Library size, bp')\n",
    "plt.grid()\n",
    "plt.xscale('log')\n",
    "plt.tight_layout()\n",
    "# plt.savefig('hifi_size_vs_numbp.pdf', format='pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "0b1bd8a9",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:30.159709Z",
     "start_time": "2021-11-09T14:14:29.951048Z"
    }
   },
   "outputs": [],
   "source": [
    "# adapted from https://gist.github.com/jquacinella/1a6341f0f1446973714c\n",
    "import weighted  # install with 'pip install wquantiles'\n",
    "from matplotlib.cbook import violin_stats\n",
    "from scipy import stats\n",
    "from scipy.stats import gaussian_kde\n",
    "\n",
    "\n",
    "def vdensity_with_weights(weights):\n",
    "    ''' Outer function allows innder function access to weights. Matplotlib\n",
    "    needs function to take in data and coords, so this seems like only way\n",
    "    to 'pass' custom density function a set of weights '''\n",
    "\n",
    "    def vdensity(data, coords):\n",
    "        ''' Custom matplotlib weighted violin stats function '''\n",
    "        # Using weights from closure, get KDE fomr statsmodels\n",
    "        weighted_cost = gaussian_kde(data, weights=weights)\n",
    "\n",
    "        # Return y-values for graph of KDE by evaluating on coords\n",
    "        return weighted_cost.evaluate(coords)\n",
    "    return vdensity\n",
    "\n",
    "def custom_violin(data, weights, ax, positions=None, drop_outliers=False, **kwargs):\n",
    "    results = []\n",
    "\n",
    "    for i, (x, w) in enumerate(zip(data, weights)):\n",
    "        idx = np.ones(x.shape, dtype=bool)\n",
    "        if drop_outliers:\n",
    "            q1 = weighted.quantile_1D(x, w, 0.25)\n",
    "            q3 = weighted.quantile_1D(x, w, 0.75)\n",
    "            iqr = q3 - q1  # Interquartile range\n",
    "            fence_low = q1 - (1.5 * iqr)\n",
    "            fence_high = q3 + (1.5 * iqr)\n",
    "            idx = (x > fence_low) & (x < fence_high)\n",
    "\n",
    "        # Use matplotlib violin_stats, which expects a function that takes in data and coords\n",
    "        # which we get from closure above\n",
    "        results.append(violin_stats(np.array(x)[idx], vdensity_with_weights(np.array(w)[idx].astype(float)))[0])\n",
    "\n",
    "        # Get weighted median and mean (using weighted module for median)\n",
    "        median = weighted.quantile_1D(x, w, 0.5)\n",
    "        mean, sumw = np.ma.average(x, weights=w, returned=True)\n",
    "\n",
    "        # Update result dictionary with our updated info\n",
    "        results[i][u\"mean\"] = mean\n",
    "        results[i][u\"median\"] = median\n",
    "\n",
    "        # No need to do this, since it should be populated from violin_stats\n",
    "        # results[i][u\"min\"] =  np.min(x)\n",
    "        # results[i][u\"max\"] =  np.max(x)\n",
    "\n",
    "        out = x[~idx]\n",
    "        # if there are too many points in outliers because of the size of the\n",
    "        # dataset, we sample some of them to speed up the rendering of the plot\n",
    "        T = 0.003\n",
    "        if out.size > 1000:\n",
    "            out = np.sort(out)\n",
    "            last = out[0]\n",
    "            l = [last]\n",
    "            for v in out[1:]:\n",
    "                if v - last > T:\n",
    "                    last = v\n",
    "                    l.append(last)\n",
    "            out = np.array(l)\n",
    "            print(\"Num outliers rendered: {}\".format(out.size))\n",
    "\n",
    "        if 'vert' in kwargs.keys() and kwargs['vert']:\n",
    "            ax.scatter(positions[i] * np.ones(out.shape), out,\n",
    "                       marker='o', linewidth=0.3, edgecolor='black', color='white', s=3, zorder=3)\n",
    "        else:\n",
    "            ax.scatter(out, positions[i] * np.ones(out.shape),\n",
    "                       marker='o', linewidth=0.3, edgecolor='black', color='white', s=3, zorder=3)\n",
    "\n",
    "    parts = ax.violin(results, positions=positions, widths=0.6, **kwargs)\n",
    "    for partname in ('cmeans',):\n",
    "        parts[partname].set_colors('black')\n",
    "    return parts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "f3bba44b",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:32.165224Z",
     "start_time": "2021-11-09T14:14:30.162344Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Num outliers rendered: 528\n",
      "Num outliers rendered: 285\n",
      "Num outliers rendered: 223\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 376.596x155.169 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "width = 3.487 * 1.5 * 1\n",
    "height = width / 1.618 / 1.5\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(width, height))\n",
    "\n",
    "violin_data = [\n",
    "    8 * spring_size / num_bp,\n",
    "    8 * mtg2_size / num_bp,\n",
    "    8 * mtg_size / num_bp,\n",
    "    8 * gzip_size / num_bp,\n",
    "    8 * blast_size / num_bp,\n",
    "]\n",
    "labels = ['Spring (*)',\n",
    "          'Counting DBG with coord.\\n(multiple successors)',\n",
    "          'Counting DBG with coord.\\n(single successor)',\n",
    "          'gzip -9 (*)',\n",
    "          'BLAST',]\n",
    "violin_weights = [num_bp] * len(labels)\n",
    "pos = np.arange(len(violin_data)) / 1.3\n",
    "parts = custom_violin(violin_data, positions=pos, weights=violin_weights,\n",
    "              vert=False, showmeans=True, ax=ax, drop_outliers=True)\n",
    "#plt.violinplot(violin_data, vert=False, showmeans=True)\n",
    "\n",
    "def set_axis_style(ax, labels):\n",
    "    ax.yaxis.set_tick_params(direction='in')\n",
    "    ax.yaxis.set_ticks_position('left')\n",
    "    ax.set_yticks(pos)\n",
    "    ax.set_yticklabels(labels, size=11)\n",
    "\n",
    "set_axis_style(ax, labels)\n",
    "\n",
    "#plt.axvline(1, color='black', alpha=0.5, linestyle='dashed', linewidth=2)\n",
    "\n",
    "#plt.legend(fontsize=12)\n",
    "plt.xlabel('Size, bits/bp', size=11)\n",
    "plt.grid()\n",
    "plt.tight_layout()\n",
    "#plt.xlim([0, 15])\n",
    "plt.savefig('hifi_compression_violin.pdf', format='pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "648197ec",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:37.104674Z",
     "start_time": "2021-11-09T14:14:32.167182Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Num outliers rendered: 222\n",
      "Num outliers rendered: 187\n",
      "Num outliers rendered: 216\n",
      "Num outliers rendered: 150\n",
      "Num outliers rendered: 82\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 753.192x258.616 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "width = 3.487 * 1.5 * 2\n",
    "height = width / 1.618 / 1.8\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(width, height))\n",
    "\n",
    "colors = ['b', 'orange', 'green', 'red']\n",
    "labels = ['gzip -9 (*)',\n",
    "          'Spring (*)',\n",
    "          'BLAST',\n",
    "          'This work']\n",
    "\n",
    "bs = [1e5, 1e6, 1e7, 1e8, np.inf]\n",
    "num_points = []\n",
    "for i in range(len(bs)):\n",
    "    if i == 0:\n",
    "        idx = num_bp < bs[0]\n",
    "    elif i == len(bs) - 1:\n",
    "        idx = num_bp > bs[-2]\n",
    "    else:\n",
    "        idx = (num_bp < bs[i]) & (num_bp >= bs[i - 1])\n",
    "    num_points.append(sum(idx))\n",
    "\n",
    "    violin_data = [gzip_size[idx] * 8 / num_bp[idx],\n",
    "                   spring_size[idx] * 8 / num_bp[idx],\n",
    "                   blast_size[idx] * 8 / num_bp[idx],\n",
    "                   mtg_size[idx] * 8 / num_bp[idx]]\n",
    "    violin_weights = [num_bp[idx],\n",
    "                      num_bp[idx],\n",
    "                      num_bp[idx],\n",
    "                      num_bp[idx],]\n",
    "    shift = len(violin_data)\n",
    "    pos = np.arange(1, len(violin_data) + 1) / 1.4\n",
    "    parts = custom_violin(violin_data, positions=shift*i + pos, weights=violin_weights,\n",
    "                          vert=True, showmeans=True, ax=ax, drop_outliers=True)\n",
    "#     parts = plt.violinplot(violin_data, positions=pos, vert=True, showmeans=True)\n",
    "#     for j in range(len(violin_data)):\n",
    "#         points = np.random.choice(violin_data[j], 10)\n",
    "#         sns.swarmplot(x=np.ones((points.size,)) * (shift*i + pos[j]), y=points,\n",
    "#                       color=\"black\", edgecolor=\"gray\", ax=ax)\n",
    "    for partname in ('cbars', 'cmins', 'cmaxes', 'cmeans'):\n",
    "        parts[partname].set_colors(colors)\n",
    "    for partname in ('cmeans',):\n",
    "        parts[partname].set_colors('black')\n",
    "#     ax.scatter(shift*i + pos, (np.array(violin_data) * np.array(violin_weights)).sum(1) / np.array(violin_weights).sum(1),\n",
    "#                marker='o', color='black', s=30, zorder=3)\n",
    "    for pc, c in zip(parts['bodies'], colors):\n",
    "        pc.set_color(c)\n",
    "        pc.set_alpha(0.3)\n",
    "        pc.set_edgecolor(c)\n",
    "        pc.set_linewidth(0.5)\n",
    "    ax.text((shift*i + pos[0] + shift*i + pos[-1]) / 2, 3.54, f'Group size: {num_points[i]}',\n",
    "            horizontalalignment='center', size=9)\n",
    "\n",
    "ax.yaxis.set_tick_params(direction='in')\n",
    "ax.yaxis.set_ticks_position('left')\n",
    "ax.xaxis.set_tick_params(length=0)\n",
    "#ax.xaxis.set_ticks_position('left')\n",
    "#ax.set_yticks(pos)\n",
    "ax.set_xticks([(shift*i + pos[0] + shift*i + pos[-1]) / 2 for i in range(len(bs))])\n",
    "# ax.set_yticks(np.arange(1, len(labels) + 1))\n",
    "def mult(x):\n",
    "    return int(x / 10**int(np.log10(x)))\n",
    "ax.set_xticklabels([f'$<10^{int(np.log10(bs[0]))}$'] +\n",
    "                   [f'$[{f\"{mult(bs[i-1])}{chr(92)}cdot \" if mult(bs[i-1]) != 1 else \"\"}10^{int(np.log10(bs[i-1]))}, '\n",
    "                    f'{f\"{mult(bs[i])}{chr(92)}cdot \" if mult(bs[i]) != 1 else \"\"}10^{int(np.log10(bs[i]))})$'\n",
    "                    for i in range(1, len(bs)-1)] +\n",
    "                   [f'$\\geq 10^{int(np.log10(bs[-2]))}$'])\n",
    "ax.xaxis.grid(True)\n",
    "ax.legend(parts['bodies'], labels, loc='lower left')\n",
    "#ax.set_ylim(0.25, len(labels) + 0.75)\n",
    "\n",
    "#plt.axvline(1, color='black', alpha=0.5, linestyle='dashed', linewidth=2)\n",
    "\n",
    "plt.ylabel('Size, bits/bp')\n",
    "plt.xlabel('Library size, bp')\n",
    "#plt.title('HiFi viruses')\n",
    "plt.grid()\n",
    "plt.tight_layout()\n",
    "#plt.xlim([0, 15])\n",
    "plt.savefig('hifi_compression_violin_groups.pdf', format='pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "feb55e84",
   "metadata": {},
   "source": [
    "#### Query time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "98242b99",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-22T13:36:54.737797Z",
     "start_time": "2022-04-22T13:36:54.714619Z"
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "id": "1d6fd607",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-26T13:10:41.760762Z",
     "start_time": "2022-04-26T13:09:19.866872Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.09, 0.51\n",
      "0.08, 2.06\n",
      "0.08, 0.11\n",
      "0.15, 0.41\n"
     ]
    }
   ],
   "source": [
    "def show_time(times_file):\n",
    "    df = pd.read_csv(times_file, header=None, sep=' ', names=['id', 'time'], index_col=0)\n",
    "    for i in range(df.shape[0]):\n",
    "        try:\n",
    "            df.iloc[i].time = to_seconds(df.iloc[i].time)\n",
    "        except:\n",
    "            print(df.iloc[i])\n",
    "            df.iloc[i].time = 0\n",
    "    print(f'{df.time.mean():.2f}, {df.time.std():.2f}')\n",
    "\n",
    "#print(show_time('~/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/mtg/delta_variants_query.times'))\n",
    "show_time('~/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/megablast/delta_variants_query.times')\n",
    "show_time('~/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/pufferfish_sparse/delta_variants_align.times')\n",
    "show_time('~/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/blast/delta_variants_query.times')\n",
    "show_time('~/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/mtg/delta_variants_align.times')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "id": "31c0f6a7",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-26T13:16:29.246970Z",
     "start_time": "2022-04-26T13:15:07.360418Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.09, 0.49\n",
      "0.06, 1.06\n",
      "0.08, 0.10\n",
      "0.15, 0.40\n"
     ]
    }
   ],
   "source": [
    "def show_time(times_file):\n",
    "    df = pd.read_csv(times_file, header=None, sep=' ', names=['id', 'time'], index_col=0)\n",
    "    df = df.loc[np.array([x.split('/')[-1][:-len('_no_header.fasta.gz')] for x in gzip])]\n",
    "    for i in range(df.shape[0]):\n",
    "        try:\n",
    "            df.iloc[i].time = to_seconds(df.iloc[i].time)\n",
    "        except:\n",
    "            print(df.iloc[i])\n",
    "            df.iloc[i].time = 0\n",
    "    print(f'{df.time.mean():.2f}, {df.time.std():.2f}')\n",
    "\n",
    "#show_time('~/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/mtg/delta_variants_query.times')\n",
    "show_time('~/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/megablast/delta_variants_query.times')\n",
    "show_time('~/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/pufferfish_sparse/delta_variants_align.times')\n",
    "show_time('~/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/blast/delta_variants_query.times')\n",
    "show_time('~/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/mtg/delta_variants_align.times')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "dc2bfb02",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-22T14:33:47.588583Z",
     "start_time": "2022-04-22T14:33:47.431837Z"
    }
   },
   "outputs": [],
   "source": [
    "df = pd.read_csv('~/metagenome/finished_projects/counting_dbg/hifi_sra/viruses_hifi_data/mtg/delta_variants_align_chain_optim.times', header=None, sep=' ', names=['id', 'time'], index_col=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "92903856",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-22T14:34:52.219848Z",
     "start_time": "2022-04-22T14:34:32.462326Z"
    }
   },
   "outputs": [],
   "source": [
    "for i in range(df.shape[0]):\n",
    "    df.iloc[i].time = to_seconds(df.iloc[i].time)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "a691b9a5",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-22T14:35:05.536232Z",
     "start_time": "2022-04-22T14:35:05.499065Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>SRR13144527</th>\n",
       "      <td>2133.2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               time\n",
       "id                 \n",
       "SRR13144527  2133.2"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df.time == 2133.2]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0720b2d6",
   "metadata": {},
   "source": [
    "### Align to RefSeq"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "06c11e73",
   "metadata": {},
   "source": [
    "#### Fungi"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2da4c496",
   "metadata": {},
   "source": [
    "BLAST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "id": "94d6597f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-24T16:03:53.380371Z",
     "start_time": "2022-04-24T16:03:53.286701Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Runs: 100\n",
      "2.07 sec\n",
      "2.175 GB\n"
     ]
    }
   ],
   "source": [
    "FILE = '/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_try2/SRR10002688_1.shuf.head100.split.fungi.blast.log'\n",
    "\n",
    "lines = !grep 'wall' $FILE\n",
    "\n",
    "print('Runs:', len(lines))\n",
    "print('{:.2f} sec'.format(np.mean([to_seconds(x.split(' ')[-1]) for x in lines])))\n",
    "\n",
    "lines = !grep 'Maximum resident set size' $FILE\n",
    "\n",
    "print('{:.3f} GB'.format(np.mean([float(x.split(' ')[-1]) / 1e6 for x in lines])))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "24ea90f0",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-24T16:03:29.832878Z",
     "start_time": "2022-04-24T16:03:29.403798Z"
    }
   },
   "source": [
    "MegaBLAST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "id": "2a859bd3",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-24T16:03:58.523237Z",
     "start_time": "2022-04-24T16:03:58.441588Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Runs: 100\n",
      "4.31 sec\n",
      "0.034 GB\n"
     ]
    }
   ],
   "source": [
    "FILE = '/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_try2/SRR10002688_1.shuf.head100.split.fungi.megablast.log'\n",
    "\n",
    "lines = !grep 'wall' $FILE\n",
    "\n",
    "print('Runs:', len(lines))\n",
    "print('{:.2f} sec'.format(np.mean([to_seconds(x.split(' ')[-1]) for x in lines])))\n",
    "\n",
    "lines = !grep 'Maximum resident set size' $FILE\n",
    "\n",
    "print('{:.3f} GB'.format(np.mean([float(x.split(' ')[-1]) / 1e6 for x in lines])))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "65da4202",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-24T16:05:28.998442Z",
     "start_time": "2022-04-24T16:05:28.513571Z"
    }
   },
   "source": [
    "MetaGraph"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 226,
   "id": "2b8e817e",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-27T08:36:59.711362Z",
     "start_time": "2022-04-27T08:36:59.617953Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Runs: 100\n",
      "0.021 sec\n",
      "3.4 GB\n"
     ]
    }
   ],
   "source": [
    "FILE = '/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_mtg_final/SRR10002688_1.shuf.head100.split.fungi.align.log'\n",
    "\n",
    "lines = !grep 'processed in' $FILE\n",
    "\n",
    "print('Runs:', len(lines))\n",
    "print('{:.3f} sec'.format(np.mean([float(x.split(' ')[7]) for x in lines])))\n",
    "\n",
    "lines = !grep 'Maximum resident set size' $FILE\n",
    "print('{:.1f} GB'.format(np.mean([float(x.split(' ')[-1]) / 1e6 for x in lines])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "id": "13a32580",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-27T08:37:45.259072Z",
     "start_time": "2022-04-27T08:37:44.280431Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Runs: 1\n",
      "14.76 sec\n",
      "3.5 GB\n"
     ]
    }
   ],
   "source": [
    "FILE = '/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_mtg_final/SRR10002688_1.shuf.head1000.fa.fungi.align.log'\n",
    "\n",
    "lines = !grep 'processed in' $FILE\n",
    "\n",
    "print('Runs:', len(lines))\n",
    "print('{:.2f} sec'.format(np.mean([float(x.split(' ')[7]) for x in lines])))\n",
    "\n",
    "lines = !grep 'Maximum resident set size' $FILE\n",
    "print('{:.1f} GB'.format(np.mean([float(x.split(' ')[-1]) / 1e6 for x in lines])))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8674e8ec",
   "metadata": {},
   "source": [
    "#### All"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "81ef66ec",
   "metadata": {},
   "source": [
    "BLAST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "id": "26f9f305",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-27T07:37:48.430567Z",
     "start_time": "2022-04-27T07:37:48.337182Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Runs: 100\n",
      "353 sec\n",
      "417 GB\n"
     ]
    }
   ],
   "source": [
    "FILES = '/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_try3/nobackup/SRR10002688_1.shuf.head1000.split.SRR10002688_1.shuf.head1000.*.fa.blast.log'\n",
    "\n",
    "lines = !grep 'wall' $FILES\n",
    "\n",
    "print('Runs:', len(lines))\n",
    "print('{:.0f} sec'.format(np.mean([to_seconds(x.split(' ')[-1]) for x in lines])))\n",
    "\n",
    "lines = !grep 'Maximum resident set size' $FILES\n",
    "\n",
    "print('{:.0f} GB'.format(np.mean([float(x.split(' ')[-1]) / 1e6 for x in lines])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "id": "6da9777f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-26T08:58:01.674094Z",
     "start_time": "2022-04-26T08:58:01.592426Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Runs: 1\n",
      "1857 sec\n",
      "428 GB\n"
     ]
    }
   ],
   "source": [
    "FILE = '/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_try2/SRR10002688_1.shuf.head1000.fa.refseq.blast.log'\n",
    "\n",
    "lines = !grep 'wall' $FILE\n",
    "\n",
    "print('Runs:', len(lines))\n",
    "print('{:.0f} sec'.format(np.mean([to_seconds(x.split(' ')[-1]) for x in lines])))\n",
    "\n",
    "lines = !grep 'Maximum resident set size' $FILE\n",
    "\n",
    "print('{:.0f} GB'.format(np.mean([float(x.split(' ')[-1]) / 1e6 for x in lines])))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2751f9a0",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-24T16:03:29.832878Z",
     "start_time": "2022-04-24T16:03:29.403798Z"
    }
   },
   "source": [
    "MegaBLAST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "id": "72680fa3",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-27T07:29:01.194861Z",
     "start_time": "2022-04-27T07:28:50.008327Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Runs: 100\n",
      "Runs: 100\n",
      "Runs: 100\n",
      "Runs: 100\n",
      "Runs: 100\n",
      "12.5 sec\n",
      "0.090 GB\n"
     ]
    }
   ],
   "source": [
    "times = []\n",
    "rams = []\n",
    "\n",
    "for i in [0, 1, 2, 3, 4]:\n",
    "    prefix = 'fatnode_try3' if i >= 2 else 'fatnode_try4'\n",
    "    FILES = f'/cluster/work/grlab/projects/metagenome/data/refseq/queries/{prefix}/nobackup/refseq_{i}.SRR10002688_1.shuf.head1000.split.SRR10002688_1.shuf.head1000.*.fa.megablast.log'\n",
    "\n",
    "    lines = !grep 'wall' $FILES\n",
    "    print('Runs:', len(lines))\n",
    "\n",
    "    times.append(np.mean([to_seconds(x.split(' ')[-1]) for x in lines]))\n",
    "\n",
    "    lines = !grep 'Maximum resident set size' $FILES\n",
    "    rams.append(np.mean([float(x.split(' ')[-1]) / 1e6 for x in lines]))\n",
    "\n",
    "print('{:.1f} sec'.format(np.sum(times)))\n",
    "print('{:.3f} GB'.format(np.max(rams)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 216,
   "id": "4e300d36",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-27T07:37:48.430567Z",
     "start_time": "2022-04-27T07:37:48.337182Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_try3/nobackup/refseq_2.SRR10002688_1.shuf.head1000.SRR10002688_1.shuf.head1000.fa.megablast.log:\tElapsed (wall clock) time (h:mm:ss or m:ss): 2:52.64\n",
      "/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_try3/nobackup/refseq_3.SRR10002688_1.shuf.head1000.SRR10002688_1.shuf.head1000.fa.megablast.log:\tElapsed (wall clock) time (h:mm:ss or m:ss): 6:21.01\n",
      "/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_try3/nobackup/refseq_4.SRR10002688_1.shuf.head1000.SRR10002688_1.shuf.head1000.fa.megablast.log:\tElapsed (wall clock) time (h:mm:ss or m:ss): 3:15.47\n",
      "/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_try4/nobackup/refseq_0.SRR10002688_1.shuf.head1000.SRR10002688_1.shuf.head1000.fa.megablast.log:\tElapsed (wall clock) time (h:mm:ss or m:ss): 6:10.30\n",
      "/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_try4/nobackup/refseq_1.SRR10002688_1.shuf.head1000.SRR10002688_1.shuf.head1000.fa.megablast.log:\tElapsed (wall clock) time (h:mm:ss or m:ss): 7:02.20\n",
      "Runs: 5\n",
      "1542 sec\n",
      "22.0 GB\n"
     ]
    }
   ],
   "source": [
    "FILES = '/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_try*/nobackup/refseq_*.SRR10002688_1.shuf.head1000.SRR10002688_1.shuf.head1000.fa.megablast.log'\n",
    "\n",
    "lines = !grep 'wall' $FILES\n",
    "for x in lines:\n",
    "    print(x)\n",
    "\n",
    "print('Runs:', len(lines))\n",
    "print('{:.0f} sec'.format(np.sum([to_seconds(x.split(' ')[-1]) for x in lines])))\n",
    "\n",
    "lines = !grep 'Maximum resident set size' $FILES\n",
    "\n",
    "print('{:.1f} GB'.format(np.max([float(x.split(' ')[-1]) / 1e6 for x in lines])))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "26b7b7b3",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-24T16:05:28.998442Z",
     "start_time": "2022-04-24T16:05:28.513571Z"
    }
   },
   "source": [
    "MetaGraph"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "id": "671888f0",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-27T08:34:15.332054Z",
     "start_time": "2022-04-27T08:34:15.250214Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Runs: 100\n",
      "0.66 sec\n",
      "500 GB\n"
     ]
    }
   ],
   "source": [
    "FILE = '/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_mtg_final/SRR10002688_1.shuf.head100.split.refseq.align.log'\n",
    "\n",
    "lines = !grep 'processed in' $FILE\n",
    "\n",
    "print('Runs:', len(lines))\n",
    "print('{:.2f} sec'.format(np.mean([float(x.split(' ')[7]) for x in lines])))\n",
    "\n",
    "lines = !grep 'Maximum resident set size' $FILE\n",
    "print('{:.0f} GB'.format(np.mean([float(x.split(' ')[-1]) / 1e6 for x in lines])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "id": "56c3071f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-04-27T08:33:30.383423Z",
     "start_time": "2022-04-27T08:33:30.249041Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Runs: 1\n",
      "575 sec\n",
      "513 GB\n"
     ]
    }
   ],
   "source": [
    "FILE = '/cluster/work/grlab/projects/metagenome/data/refseq/queries/fatnode_mtg_final/SRR10002688_1.shuf.head1000.fa.refseq.align.log'\n",
    "\n",
    "lines = !grep 'processed in' $FILE\n",
    "\n",
    "print('Runs:', len(lines))\n",
    "print('{:.0f} sec'.format(np.mean([float(x.split(' ')[7]) for x in lines])))\n",
    "\n",
    "lines = !grep 'Maximum resident set size' $FILE\n",
    "print('{:.0f} GB'.format(np.mean([float(x.split(' ')[-1]) / 1e6 for x in lines])))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "03209b5d",
   "metadata": {},
   "source": [
    "## Illumina RNA-Seq reads"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "3d872be3",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-01-17T19:59:44.156468Z",
     "start_time": "2022-01-17T19:59:43.671793Z"
    }
   },
   "outputs": [],
   "source": [
    "with open('kingsford_gzip_spring.pkl', 'rb') as f:\n",
    "    gzip, read_length, num_bp, gzip_size, fasta_size, spring_size = pickle.load(f)\n",
    "\n",
    "# gzip = !ls ~/metagenome/data/kingsford/compressed/*_no_header.fasta.gz\n",
    "\n",
    "# num_bp = []\n",
    "# read_length = []\n",
    "# spring_size = []\n",
    "# gzip_size = []\n",
    "# fasta_size = []\n",
    "\n",
    "# for x in tqdm(gzip, mininterval=1):\n",
    "#     gzip_size.append(get_size(x))\n",
    "#     fasta_size.append(get_size(x[:-len('.gz')]))\n",
    "#     ID = x[:-len('_no_header.fasta.gz')]\n",
    "#     spring_fname = f'{ID}_no_header.spring'\n",
    "#     spring_size.append(get_size(spring_fname))\n",
    "#     SRA = ID.split('/')[-1]\n",
    "#     fname = x[:-len('_no_header.fasta.gz')] + '.num_bp'\n",
    "#     with open(fname, 'r') as f:\n",
    "#         n = int(f.readline())\n",
    "#         num_bp.append(n)\n",
    "\n",
    "#     with gz.open(x, 'r') as f:\n",
    "#         f.readline().strip().decode()\n",
    "#         seq = f.readline().strip().decode()\n",
    "#         L = len(seq)\n",
    "#         all_same_len = True\n",
    "#         for i in range(1000):  # check 1000 first reads\n",
    "#             f.readline().strip().decode()\n",
    "#             if L != len(f.readline().strip().decode()):\n",
    "#                 all_same_len = False\n",
    "#                 break\n",
    "#         read_length.append(L if all_same_len and num_bp[-1] % L == 0 else np.nan)\n",
    "\n",
    "# gzip = np.array(gzip)\n",
    "# num_bp = np.array(num_bp)\n",
    "# spring_size = np.array(spring_size)\n",
    "# gzip_size = np.array(gzip_size)\n",
    "# fasta_size = np.array(fasta_size)\n",
    "# read_length = np.array(read_length)\n",
    "\n",
    "# with open('kingsford_gzip_spring.pkl', 'wb') as f:\n",
    "#     pickle.dump((gzip, read_length, num_bp, gzip_size, fasta_size, spring_size), f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "4ffcdd65",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-01-17T19:59:46.692236Z",
     "start_time": "2022-01-17T19:59:46.667700Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2652, 7.973392066664)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(num_bp), sum(num_bp) / 1e12"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "cae57556",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-01-17T19:59:49.473344Z",
     "start_time": "2022-01-17T19:59:49.448238Z"
    }
   },
   "outputs": [],
   "source": [
    "with open('kingsford_mtg_size.pkl', 'rb') as f:\n",
    "    mtg_size = pickle.load(f)\n",
    "\n",
    "# mtg_size = []\n",
    "\n",
    "# DIR = !echo $HOME\n",
    "# DIR = f'{DIR[0]}/metagenome/data/kingsford_31_coordinates/'\n",
    "\n",
    "# for x in tqdm(gzip, mininterval=1):\n",
    "#     ID = x.split('/')[-1][:-len('_no_header.fasta.gz')]\n",
    "#     graph_size = get_size(f'{DIR}/{ID}.fasta.gz/graph_small.dbg')\n",
    "#     graph_size += get_size(f'{DIR}/{ID}.fasta.gz/graph.dbg.rd_succ')\n",
    "#     graph_size += get_size(f'{DIR}/{ID}.fasta.gz/graph.dbg.anchors')\n",
    "#     column_fname = f'{DIR}/{ID}.fasta.gz/rd_columns/annotation.column.annodbg'\n",
    "#     coords_fname = f'{DIR}/{ID}.fasta.gz/rd_columns/annotation.column.annodbg.coords'\n",
    "#     mtg_size.append(graph_size + get_size(column_fname) + get_size(coords_fname))\n",
    "\n",
    "# mtg_size = np.array(mtg_size)\n",
    "\n",
    "# with open('kingsford_mtg_size.pkl', 'wb') as f:\n",
    "#     pickle.dump(mtg_size, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "09188d87",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-01-17T20:08:08.444781Z",
     "start_time": "2022-01-17T20:08:08.421148Z"
    },
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "with open('kingsford_mtg_2_size.pkl', 'rb') as f:\n",
    "    mtg2_size = pickle.load(f)\n",
    "\n",
    "# mtg2_size = []\n",
    "\n",
    "# DIR = !echo $HOME\n",
    "# DIR = f'{DIR[0]}/metagenome/finished_projects/counting_dbg/kingsford_31_coordinates_fork_opt_new'\n",
    "\n",
    "# for x in tqdm(gzip, mininterval=1):\n",
    "#     ID = x.split('/')[-1][:-len('_no_header.fasta.gz')]\n",
    "#     graph_size = get_size(f'{DIR}/../kingsford_31_coordinates/{ID}.fasta.gz/graph_small.dbg')\n",
    "#     graph_size += get_size(f'{DIR}/{ID}.fasta.gz/graph.dbg.rd_succ')\n",
    "#     graph_size += get_size(f'{DIR}/{ID}.fasta.gz/graph.dbg.anchors')\n",
    "#     column_fname = f'{DIR}/{ID}.fasta.gz/rd_columns/annotation.column.annodbg'\n",
    "#     coords_fname = f'{DIR}/{ID}.fasta.gz/rd_columns/annotation.column.annodbg.coords'\n",
    "#     mtg2_size.append(graph_size + get_size(column_fname) + get_size(coords_fname))\n",
    "\n",
    "# mtg2_size = np.array(mtg2_size)\n",
    "\n",
    "# with open('kingsford_mtg_2_size.pkl', 'wb') as f:\n",
    "#     pickle.dump(mtg2_size, f)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e749d076",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-08T11:22:22.293017Z",
     "start_time": "2021-11-08T11:22:22.269550Z"
    }
   },
   "source": [
    "Select sets of reads of fixed length (31 or longer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "fc780c32",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:37.216089Z",
     "start_time": "2021-11-09T14:14:37.202692Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2411 read sets (90.9%), 7.55 Tbp\n"
     ]
    }
   ],
   "source": [
    "idx = (read_length >= 31)\n",
    "print(f'{idx.sum()} read sets ({idx.mean()*100:.1f}%), {num_bp[idx].sum() / 1e12:.2f} Tbp')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3b39225c",
   "metadata": {},
   "source": [
    "Excluded:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "32ef38e9",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:37.238081Z",
     "start_time": "2021-11-09T14:14:37.218259Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['SRR037695', 'SRR037703', 'SRR037704', 'SRR037705', 'SRR037706',\n",
       "       'SRR037707', 'SRR037708', 'SRR037709', 'SRR060885', 'SRR060886',\n",
       "       'SRR060887', 'SRR060888', 'SRR060889', 'SRR060890', 'SRR060891',\n",
       "       'SRR060892', 'SRR060893', 'SRR060894', 'SRR060895', 'SRR060896',\n",
       "       'SRR060897', 'SRR060898', 'SRR060899', 'SRR060900', 'SRR060901',\n",
       "       'SRR060902', 'SRR060903', 'SRR060904', 'SRR060905', 'SRR060906',\n",
       "       'SRR060907', 'SRR060908', 'SRR060909', 'SRR060910', 'SRR060911',\n",
       "       'SRR067891', 'SRR067892', 'SRR067893', 'SRR067894', 'SRR067895',\n",
       "       'SRR067896', 'SRR067897', 'SRR067898', 'SRR067899', 'SRR067900',\n",
       "       'SRR067901', 'SRR067902', 'SRR067903', 'SRR067904', 'SRR079377',\n",
       "       'SRR079378', 'SRR079379', 'SRR079380', 'SRR079381', 'SRR079382',\n",
       "       'SRR079383', 'SRR079384', 'SRR1283038', 'SRR1292576', 'SRR1292577',\n",
       "       'SRR1292578', 'SRR1292579', 'SRR1292580', 'SRR1292581',\n",
       "       'SRR1292582', 'SRR1292583', 'SRR1292584', 'SRR1313065',\n",
       "       'SRR1409743', 'SRR1409744', 'SRR1409745', 'SRR1409746',\n",
       "       'SRR1409747', 'SRR223506', 'SRR343343', 'SRR343344', 'SRR343345',\n",
       "       'SRR343346', 'SRR343347', 'SRR343348', 'SRR343349', 'SRR343350',\n",
       "       'SRR343351', 'SRR343352', 'SRR343353', 'SRR343354', 'SRR343355',\n",
       "       'SRR343356', 'SRR346127', 'SRR346128', 'SRR346129', 'SRR363685',\n",
       "       'SRR363686', 'SRR363687', 'SRR363688', 'SRR363689', 'SRR363690',\n",
       "       'SRR363691', 'SRR363693', 'SRR363694', 'SRR363695', 'SRR363696',\n",
       "       'SRR363699', 'SRR363700', 'SRR363701', 'SRR448330', 'SRR448331',\n",
       "       'SRR515313', 'SRR537128', 'SRR537130', 'SRR537131', 'SRR537132',\n",
       "       'SRR543502', 'SRR543503', 'SRR543504', 'SRR543505', 'SRR543506',\n",
       "       'SRR543507', 'SRR543508', 'SRR543509', 'SRR543510', 'SRR543511',\n",
       "       'SRR543512', 'SRR543513', 'SRR543514', 'SRR543515', 'SRR543516',\n",
       "       'SRR543517', 'SRR543518', 'SRR543519', 'SRR543520', 'SRR543521',\n",
       "       'SRR543522', 'SRR543523', 'SRR543524', 'SRR543525', 'SRR543526',\n",
       "       'SRR543527', 'SRR543528', 'SRR543529', 'SRR543530', 'SRR543531',\n",
       "       'SRR543532', 'SRR543533', 'SRR578243', 'SRR578244', 'SRR578245',\n",
       "       'SRR578246', 'SRR578247', 'SRR578248', 'SRR578249', 'SRR578250',\n",
       "       'SRR578251', 'SRR578252', 'SRR578253', 'SRR578254', 'SRR578255',\n",
       "       'SRR578256', 'SRR578257', 'SRR578258', 'SRR578259', 'SRR578260',\n",
       "       'SRR578261', 'SRR578262', 'SRR648404', 'SRR648405', 'SRR648406',\n",
       "       'SRR648407', 'SRR648408', 'SRR648409', 'SRR648410', 'SRR648411',\n",
       "       'SRR648412', 'SRR648413', 'SRR648414', 'SRR648415', 'SRR648416',\n",
       "       'SRR648417', 'SRR648418', 'SRR648419', 'SRR648420', 'SRR648421',\n",
       "       'SRR648422', 'SRR648423', 'SRR648424', 'SRR648425', 'SRR648426',\n",
       "       'SRR648427', 'SRR648428', 'SRR770720', 'SRR787327', 'SRR787329',\n",
       "       'SRR791052', 'SRR791054', 'SRR791056', 'SRR791057', 'SRR791058',\n",
       "       'SRR791059', 'SRR791061', 'SRR791062', 'SRR833713', 'SRR833714',\n",
       "       'SRR833715', 'SRR921934', 'SRR921935', 'SRR921936', 'SRR921937',\n",
       "       'SRR921938', 'SRR921939', 'SRR921940', 'SRR921941', 'SRR921942',\n",
       "       'SRR921943', 'SRR921944', 'SRR921945', 'SRR925718', 'SRR976225',\n",
       "       'SRR976226', 'SRR976227', 'SRR976228', 'SRR976229', 'SRR976230',\n",
       "       'SRR976231', 'SRR976232', 'SRR976233', 'SRR976234', 'SRR976235',\n",
       "       'SRR976236', 'SRR976237', 'SRR976238', 'SRR976239', 'SRR976240',\n",
       "       'SRR976241', 'SRR976242', 'SRR976243', 'SRR976244', 'SRR976245',\n",
       "       'SRR976246', 'SRR976247', 'SRR976248', 'SRR976249'], dtype='<U10')"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.array([x.split('/')[-1][:-len('_no_header.fasta.gz')] for x in gzip[~idx]])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9db31349",
   "metadata": {},
   "source": [
    "Counting de Bruijn graphs were constructed from all 2,411 libraries with reads of length at least 31."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "ea6fbfdd",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:38.306912Z",
     "start_time": "2021-11-09T14:14:37.239481Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 376.596x325.856 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "width = 3.487 * 1.5 * 1\n",
    "height = width / 1.618 * 1.4\n",
    "\n",
    "BYTES_IN_GB = 1e9\n",
    "\n",
    "x = np.array(gzip_size[idx]) / BYTES_IN_GB\n",
    "y = np.array(mtg2_size[idx]) / BYTES_IN_GB\n",
    "plt.figure(figsize=(width, height))\n",
    "w = num_bp[idx]\n",
    "plt.scatter(x, y, color='b', alpha=0.2, s=w / w.max() * 200)\n",
    "#plt.scatter(x, y, color='b', alpha=0.3)\n",
    "plt.plot([x.min(), x.max()], [x.min(), x.max()], ls=\"--\", c='red')\n",
    "\n",
    "i = (y / x)\n",
    "i[w < 1e9] = 0\n",
    "i = i.argsort()[::-1]\n",
    "n_labels = 7\n",
    "s = (y[i[:n_labels]] - x[i[:n_labels]]).argsort()\n",
    "for p in range(n_labels):\n",
    "    j = i[s[p]]\n",
    "    start, end = 9.7e9 / BYTES_IN_GB, 12.4e9 / BYTES_IN_GB\n",
    "    text_coord = 6.8e9 / BYTES_IN_GB, start + (end-start) * p/n_labels\n",
    "    plt.text(*text_coord, gzip[idx][j].split('/')[-1].split('_')[0], size=9)\n",
    "    plt.arrow(*text_coord, -text_coord[0] + x[j], -text_coord[1] + y[j], linestyle=':', alpha=0.5)\n",
    "\n",
    "i = (y / x)\n",
    "i[w < 1e9] = np.inf\n",
    "i = i.argsort()\n",
    "n_labels = 7\n",
    "s = (y[i[:n_labels]] - x[i[:n_labels]]).argsort()\n",
    "for p in range(n_labels):\n",
    "    j = i[s[p]]\n",
    "    start, end = 2.4e9 / BYTES_IN_GB, -0.3e9 / BYTES_IN_GB\n",
    "    text_coord = 8.1e9 / BYTES_IN_GB, start + (end-start) * p/n_labels\n",
    "    plt.text(*text_coord, gzip[idx][j].split('/')[-1].split('_')[0], size=9)\n",
    "    plt.arrow(*text_coord, -text_coord[0] + x[j], -text_coord[1] + y[j], linestyle=':', alpha=0.5)\n",
    "\n",
    "plt.ylabel('Index size, GB')\n",
    "plt.xlabel('gzip -9 (*), GB')\n",
    "plt.grid()\n",
    "plt.tight_layout()\n",
    "plt.savefig('size_vs_gzip.pdf', format='pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6b89eae0",
   "metadata": {},
   "source": [
    "We fit the curve by minimizing the weighted RMSE:\n",
    "$\\sum_{i} N_i^2 (\\frac{S_i}{N_i} - \\hat{s_i})^2\n",
    "=\\sum_{i} (S_i - N_i\\hat{s_i})^2$, that is, minimizing the size prediction error."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "483c74b6",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:39.197626Z",
     "start_time": "2021-11-09T14:14:38.309533Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 376.596x325.856 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "width = 3.487 * 1.5 * 1\n",
    "height = width / 1.618 * 1.4\n",
    "\n",
    "x = np.array(read_length[idx])\n",
    "y = np.array(mtg2_size[idx] * 8 / num_bp[idx])\n",
    "plt.figure(figsize=(width, height))\n",
    "w = num_bp[idx]\n",
    "plt.scatter(x, y, color='b', alpha=0.2, s=w / w.max() * 200)\n",
    "#plt.scatter(x, y, color='b', alpha=0.3)\n",
    "\n",
    "def f(x, a, b):\n",
    "    return a*x + b\n",
    "(a, b), pcov2 = curve_fit(f, x, y, sigma=1/num_bp[idx], absolute_sigma=True)\n",
    "plt.plot([x.min(), x.max()], [f(x.min(), a, b), f(x.max(), a, b)], 'black')\n",
    "\n",
    "plt.ylabel('Index size, bits/bp')\n",
    "plt.xlabel('Read length, bp')\n",
    "plt.grid()\n",
    "plt.tight_layout()\n",
    "plt.savefig('size_vs_readlength.pdf', format='pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "600625ce",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:39.743908Z",
     "start_time": "2021-11-09T14:14:39.201308Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 489.575x168.1 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "width = 3.487 * 1.5 * 1.3\n",
    "height = width / 1.618 / 1.8\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(width, height))\n",
    "\n",
    "violin_data = [8 * spring_size[idx] / num_bp[idx],\n",
    "               8 * mtg2_size[idx] / num_bp[idx],\n",
    "               8 * mtg_size[idx] / num_bp[idx],\n",
    "               8 * gzip_size[idx] / num_bp[idx],]\n",
    "labels = ['Spring (*)',\n",
    "          'Counting DBG with coord.\\n(multiple successors)',\n",
    "          'Counting DBG with coord.\\n(single successor)',\n",
    "          'gzip -9 (*)',]\n",
    "violin_weights = [num_bp[idx]] * len(labels)\n",
    "pos = np.arange(len(violin_data)) / 1.3\n",
    "custom_violin(violin_data, positions=pos, weights=violin_weights,\n",
    "              vert=False, showmeans=True, ax=ax, drop_outliers=True)\n",
    "#plt.violinplot(violin_data, vert=False, showmeans=True)\n",
    "\n",
    "ax.yaxis.set_tick_params(direction='in')\n",
    "ax.yaxis.set_ticks_position('left')\n",
    "ax.set_yticks(pos)\n",
    "ax.set_yticklabels(labels)\n",
    "\n",
    "#plt.axvline(1, color='black', alpha=0.5, linestyle='dashed', linewidth=2)\n",
    "\n",
    "#plt.legend(fontsize=12)\n",
    "plt.xlabel('Size, bits/bp')\n",
    "plt.grid()\n",
    "plt.tight_layout()\n",
    "#plt.xlim([0, 15])\n",
    "plt.savefig('compression_violin_extended.pdf', format='pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "1cc29d52",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:39.775137Z",
     "start_time": "2021-11-09T14:14:39.745768Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "num samples: 2411\n",
      "gzip size: 1917 GB\n",
      "spring size: 621 GB\n",
      "spring compression: 3.09 x\n",
      "spring compression: 68 %\n",
      "mtg size: 1544 GB\n",
      "mtg compression: 1.24 x\n",
      "mtg compression: 19 %\n",
      "mtg (multiple succ.) size: 1405 GB\n",
      "mtg (multiple succ.) compression: 1.36 x\n",
      "mtg (multiple succ.) compression: 27 %\n"
     ]
    }
   ],
   "source": [
    "print('num samples:', sum(idx))\n",
    "print(f'gzip size: {gzip_size[idx].sum() / 1e9:.0f} GB')\n",
    "\n",
    "print(f'spring size: {spring_size[idx].sum() / 1e9:.0f} GB')\n",
    "print(f'spring compression: {gzip_size[idx].sum() / spring_size[idx].sum():.2f} x')\n",
    "print(f'spring compression: {100 - spring_size[idx].sum() / gzip_size[idx].sum() * 100:.0f} %')\n",
    "\n",
    "print(f'mtg size: {mtg_size[idx].sum() / 1e9:.0f} GB')\n",
    "print(f'mtg compression: {gzip_size[idx].sum() / mtg_size[idx].sum():.2f} x')\n",
    "print(f'mtg compression: {100 - mtg_size[idx].sum() / gzip_size[idx].sum() * 100:.0f} %')\n",
    "\n",
    "print(f'mtg (multiple succ.) size: {mtg2_size[idx].sum() / 1e9:.0f} GB')\n",
    "print(f'mtg (multiple succ.) compression: {gzip_size[idx].sum() / mtg2_size[idx].sum():.2f} x')\n",
    "print(f'mtg (multiple succ.) compression: {100 - mtg2_size[idx].sum() / gzip_size[idx].sum() * 100:.0f} %')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "3bec41cd",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:39.801920Z",
     "start_time": "2021-11-09T14:14:39.778385Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gzip -9 compression: 3.9\n",
      "mtg(single) compression: 4.9\n",
      "mtg(mult.) compression: 5.4\n",
      "Spring compression: 12.2\n"
     ]
    }
   ],
   "source": [
    "print(f\"gzip -9 compression: {sum(num_bp[idx]) / sum(gzip_size[idx]):.1f}\")\n",
    "print(f\"mtg(single) compression: {sum(num_bp[idx]) / sum(mtg_size[idx]):.1f}\")\n",
    "print(f\"mtg(mult.) compression: {sum(num_bp[idx]) / sum(mtg2_size[idx]):.1f}\")\n",
    "print(f\"Spring compression: {sum(num_bp[idx]) / sum(spring_size[idx]):.1f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "3036cdbd",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:39.828031Z",
     "start_time": "2021-11-09T14:14:39.804164Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gzip -9 bits/bp: 2.030\n",
      "mtg(single) bits/bp: 1.636\n",
      "mtg(mult.) bits/bp: 1.488\n",
      "Spring bits/bp: 0.658\n"
     ]
    }
   ],
   "source": [
    "print(f\"gzip -9 bits/bp: {sum(gzip_size[idx]) * 8 / sum(num_bp[idx]):.3f}\")\n",
    "print(f\"mtg(single) bits/bp: {sum(mtg_size[idx]) * 8 / sum(num_bp[idx]):.3f}\")\n",
    "print(f\"mtg(mult.) bits/bp: {sum(mtg2_size[idx]) * 8 / sum(num_bp[idx]):.3f}\")\n",
    "print(f\"Spring bits/bp: {sum(spring_size[idx]) * 8 / sum(num_bp[idx]):.3f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9281347e",
   "metadata": {},
   "source": [
    "## Dependency on k-mer length"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "df4b9a8e",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:39.876998Z",
     "start_time": "2021-11-09T14:14:39.829916Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[19, 22, 25, 28, 31, 34, 37, 40, 43, 46, 49, 52]"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "DIR = '/cluster/home/mikhaika/metagenome/data/coordinates_K'\n",
    "\n",
    "ks = !ls $DIR\n",
    "ks = [16, 19, 22, 25, 28, 31, 34, 37, 40, 43, 46, 49, 52]\n",
    "ks = ks[1:]\n",
    "ks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "f5c6c3a6",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-11-09T14:14:40.859973Z",
     "start_time": "2021-11-09T14:14:39.880267Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "SRR805801 #1f77b4\n",
      "SRR3747284 #ff7f0e\n",
      "SRR13577847 #2ca02c\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 391.66x242.064 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "width = 3.487 * 1.56 * 1\n",
    "height = width / 1.618\n",
    "\n",
    "fig = plt.figure(figsize=(width, height))\n",
    "\n",
    "files = [\n",
    "    'SRR805801_no_header.fasta.gz',\n",
    "    'm64011_181218_235052.fastq.gz',\n",
    "    'SRR3747284_subreads.fastq.gz',\n",
    "#    'SRR3747411_subreads.fastq.gz',\n",
    "    'SRR386922_subreads.fastq.gz',\n",
    "#    'SRR4063132_subreads.fastq.gz',\n",
    "#    'SRR4235456_subreads.fastq.gz',\n",
    "#    'SRR11304401_subreads.fastq.gz',\n",
    "#    'SRR13684276.fastq.gz',\n",
    "    'SRR13577847_subreads.fastq.gz'\n",
    "]\n",
    "\n",
    "for ID in files:\n",
    "    total_size = []\n",
    "    graph_size = []\n",
    "    annotation_size = []\n",
    "    FILE = f'{DIR}/data/{ID[:-3]}_no_header.fasta.gz'\n",
    "    gzip_size = get_size(FILE)\n",
    "    #!zcat {FILE} | paste - - | cut -f2 | tr -d '\\n' | wc -c > {FILE}.num_bp\n",
    "    num_bp = !cat {FILE}.num_bp\n",
    "    num_bp = int(num_bp[0])\n",
    "    for k in ks:\n",
    "        g = get_size(f'{DIR}/{k}/{ID}/graph_small.dbg')\n",
    "        g += get_size(f'{DIR}/{k}/{ID}/graph.dbg.rd_succ')\n",
    "        g += get_size(f'{DIR}/{k}/{ID}/graph.dbg.anchors')\n",
    "        column_fname = f'{DIR}/{k}/{ID}/rd_columns/annotation.column.annodbg'\n",
    "        coords_fname = f'{DIR}/{k}/{ID}/rd_columns/annotation.column.annodbg.coords'\n",
    "        graph_size.append(g)\n",
    "        annotation_size.append(get_size(column_fname) + get_size(coords_fname))\n",
    "        total_size.append(graph_size[-1] + annotation_size[-1])\n",
    "\n",
    "    if np.isnan(np.array(total_size)).sum():\n",
    "        continue\n",
    "\n",
    "    p = plt.plot(ks, np.array(graph_size) * 8 / num_bp, label=f\"{ID.split('_')[0]}: Graph\", linestyle='-.', marker='')\n",
    "    plt.plot(ks, np.array(annotation_size) * 8 / num_bp, label=f\"{ID.split('_')[0]}: Annotation\", linestyle=':', marker='^', color=p[0].get_color())\n",
    "    plt.plot(ks, np.array(total_size) * 8 / num_bp, label=f\"{ID.split('_')[0]}: Total\", color=p[0].get_color())\n",
    "    print(ID.split('_')[0], p[0].get_color())\n",
    "\n",
    "ax = fig.axes[0]\n",
    "x_lim = ax.get_xlim()\n",
    "\n",
    "ax.set_xlim(x_lim)\n",
    "\n",
    "plt.xlabel('k-mer length')\n",
    "plt.ylabel('Size, bits/bp')\n",
    "plt.grid()\n",
    "\n",
    "ds = [2]\n",
    "dc = [2.2]\n",
    "asim = [2.1]\n",
    "ac = [2.1]\n",
    "\n",
    "categories = ['Graph', 'Annotation', 'Total']\n",
    "\n",
    "p11, = plt.plot([1], '-.', color='#1f77b4')\n",
    "p12, = plt.plot([1.2], ':^', color='#1f77b4')\n",
    "p13, = plt.plot([1.1], '-', color='#1f77b4')\n",
    "p21, = plt.plot([1], '-.', color='#ff7f0e')\n",
    "p22, = plt.plot([1.2], ':^', color='#ff7f0e')\n",
    "p23, = plt.plot([1.1], '-', color='#ff7f0e')\n",
    "p31, = plt.plot([1], '-.', color='#2ca02c')\n",
    "p32, = plt.plot([1.2], ':^', color='#2ca02c')\n",
    "p33, = plt.plot([1.1], '-', color='#2ca02c')\n",
    "p1, = plt.plot([0], marker='None', linestyle='None')\n",
    "p2, = plt.plot([0],  marker='None', linestyle='None')\n",
    "p3, = plt.plot([0],  marker='None', linestyle='None')\n",
    "\n",
    "leg3 = plt.legend([p1, p11, p12, p13, p2, p21, p22, p23, p3, p31, p32, p33],\n",
    "              ['SRR805801'] + categories + ['SRR3747284'] + categories + ['SRR13577847'] + categories,\n",
    "              loc='best', ncol=3, fontsize=9) # Two columns, vertical group labels\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.savefig('size_vs_k.pdf', format='pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e098b928",
   "metadata": {},
   "source": [
    "## Construction time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "041a2243",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-07T18:06:34.192082Z",
     "start_time": "2022-03-07T18:06:33.727756Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Binary OLD:\n",
      "KMC counting: 115.96 h, in 2652 runs, each using 1 cores\tNormalized time: 3.221 h \tmem: 41.455688 GB\n",
      "Build contigs: 24.58 h, in 5304 runs, each using 1 cores\tNormalized time: 0.683 h \tmem: 50.69142 GB\n",
      "Build joint graph: 1.11 h, in 4 runs, each using 36 cores\tNormalized time: 1.115 h \tmem: 51.482024 GB\n",
      "Annotate graph: 1.11 h, in 10 runs, each using 18 cores\t\tNormalized time: 0.555 h \tmem: 51.589428 GB\n",
      "Transform anno: 1.34 h, in 4 runs, each using 36 cores\t\tNormalized time: 1.337 h \tmem: 58.220948 GB\n"
     ]
    }
   ],
   "source": [
    "print('Binary OLD:')\n",
    "\n",
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/kmc_count.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/kmc_count.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'KMC counting: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/build_single.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/build_single.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Build contigs: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 36\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/build_graph.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/build_graph.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Build joint graph: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 18\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/annotate_*.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/annotate_*.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Annotate graph: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 36\n",
    "time = !grep Elapsed ~/metagenome/finished_projects/counting_dbg/kingsford_21/logs/old_rd_* | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/finished_projects/counting_dbg/kingsford_21/logs/old_rd_* | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Transform anno: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "31ea264e",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-07T18:06:50.077412Z",
     "start_time": "2022-03-07T18:06:49.740832Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Binary:\n",
      "KMC counting: 115.96 h, in 2652 runs, each using 1 cores\tNormalized time: 3.221 h \tmem: 41.455688 GB\n",
      "Build contigs: 24.58 h, in 5304 runs, each using 1 cores\tNormalized time: 0.683 h \tmem: 50.69142 GB\n",
      "Build joint graph: 1.11 h, in 4 runs, each using 36 cores\tNormalized time: 1.115 h \tmem: 51.482024 GB\n",
      "Annotate graph: 1.11 h, in 10 runs, each using 18 cores\t\tNormalized time: 0.555 h \tmem: 51.589428 GB\n",
      "Transform anno: 1.19 h, in 5 runs, each using 36 cores\t\tNormalized time: 1.191 h \tmem: 59.31488 GB\n"
     ]
    }
   ],
   "source": [
    "print('Binary:')\n",
    "\n",
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/kmc_count.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/kmc_count.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'KMC counting: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/build_single.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/build_single.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Build contigs: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 36\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/build_graph.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/build_graph.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Build joint graph: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 18\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/annotate_*.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/annotate_*.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Annotate graph: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 36\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/rd_brwt.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/rd_brwt.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Transform anno: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "609cb85a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-07T18:07:12.041814Z",
     "start_time": "2022-03-07T18:07:10.691302Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Smooth counts:\n",
      "KMC counting: 115.96 h, in 2652 runs, each using 1 cores\tNormalized time: 3.221 h \tmem: 41.455688 GB\n",
      "Build contigs: 31.10 h, in 5304 runs, each using 1 cores\tNormalized time: 0.864 h \tmem: 70.273636 GB\n",
      "Build joint graph: 1.11 h, in 4 runs, each using 36 cores\tNormalized time: 1.115 h \tmem: 51.482024 GB\n",
      "Annotate graph: 2.16 h, in 10 runs, each using 18 cores\t\tNormalized time: 1.080 h \tmem: 51.294724 GB\n",
      "Transform anno: 1.26 h, in 5 runs, each using 36 cores\t\tNormalized time: 1.256 h \tmem: 87.648508 GB\n"
     ]
    }
   ],
   "source": [
    "print('Smooth counts:')\n",
    "\n",
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/kmc_count.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/kmc_count.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'KMC counting: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/smoothing_1000000000/logs/build_single.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/smoothing_1000000000/logs/build_single.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Build contigs: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 36\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/build_graph.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/build_graph.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Build joint graph: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 18\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/smoothing_1000000000/logs/annotate_*.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/smoothing_1000000000/logs/annotate_*.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Annotate graph: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 36\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/smoothing_1000000000/logs/count_rd_brwt.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/smoothing_1000000000/logs/count_rd_brwt.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Transform anno: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "9d9db1da",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-07T18:07:40.094671Z",
     "start_time": "2022-03-07T18:07:38.708276Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Raw counts:\n",
      "KMC counting: 115.96 h, in 2652 runs, each using 1 cores\tNormalized time: 3.221 h \tmem: 41.455688 GB\n",
      "Build contigs: 29.86 h, in 5304 runs, each using 1 cores\tNormalized time: 0.829 h \tmem: 70.3226 GB\n",
      "Build joint graph: 1.11 h, in 4 runs, each using 36 cores\tNormalized time: 1.115 h \tmem: 51.482024 GB\n",
      "Annotate graph: 2.36 h, in 10 runs, each using 18 cores\t\tNormalized time: 1.179 h \tmem: 51.318896 GB\n",
      "Transform anno: 1.40 h, in 5 runs, each using 36 cores\t\tNormalized time: 1.403 h \tmem: 87.64346 GB\n"
     ]
    }
   ],
   "source": [
    "print('Raw counts:')\n",
    "\n",
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/kmc_count.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/kmc_count.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'KMC counting: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/smoothing_1/logs/build_single.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/smoothing_1/logs/build_single.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Build contigs: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 36\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/logs/build_graph.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/logs/build_graph.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Build joint graph: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 18\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/smoothing_1/logs/annotate_*.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/smoothing_1/logs/annotate_*.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Annotate graph: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')\n",
    "\n",
    "cores = 36\n",
    "time = !grep Elapsed ~/metagenome/data/kingsford_21/smoothing_1/logs/count_rd_brwt.lsf | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/data/kingsford_21/smoothing_1/logs/count_rd_brwt.lsf | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'Transform anno: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t\\tNormalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "id": "c0c15010",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-02-01T11:42:34.923564Z",
     "start_time": "2022-02-01T11:41:36.035222Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ntCard estimation (for Mantis): 72.77 h, in 2652 runs, each using 1 cores\tNormalized time: 2.021 h \tmem: 19.002412 GB\n"
     ]
    }
   ],
   "source": [
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/finished_projects/counting_dbg/kingsford_21/mantis/nobackup/logs_reruns/*k21.ntcard.4.log | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/finished_projects/counting_dbg/kingsford_21/mantis/nobackup/logs_reruns/*k21.ntcard.4.log | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'ntCard estimation (for Mantis): {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t'\n",
    "      f'Normalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "id": "4963e7de",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-02-01T12:11:24.741680Z",
     "start_time": "2022-02-01T12:11:14.463792Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CQF counting (for Mantis): 1274.37 h, in 2652 runs, each using 1 cores\tNormalized time: 35.399 h \tmem: 333.179748 GB\n"
     ]
    }
   ],
   "source": [
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/finished_projects/counting_dbg/kingsford_21/mantis/nobackup/logs_reruns/*k21.4.stderr.log | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/finished_projects/counting_dbg/kingsford_21/mantis/nobackup/logs_reruns/*k21.4.stderr.log | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'CQF counting (for Mantis): {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t'\n",
    "      f'Normalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "id": "bb2dc09c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-02-01T12:11:42.846687Z",
     "start_time": "2022-02-01T12:11:38.921845Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CQF counting (for Mantis) with 1 thread: 695.35 h, in 2652 runs, each using 1 cores\tNormalized time: 19.315 h \tmem: 482.870516 GB\n"
     ]
    }
   ],
   "source": [
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/finished_projects/counting_dbg/kingsford_21/mantis/nobackup/logs_reruns/*k21.2.stderr.log | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' ~/metagenome/finished_projects/counting_dbg/kingsford_21/mantis/nobackup/logs_reruns/*k21.2.stderr.log | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'CQF counting (for Mantis) with 1 thread: {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t'\n",
    "      f'Normalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "67dcc514",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-08T09:13:40.212561Z",
     "start_time": "2022-03-08T09:13:24.120477Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "BCALM (for REINDEER): 1462.83 h, in 2652 runs, each using 1 cores\tNormalized time: 40.634 h \tmem: 152.297732 GB\n"
     ]
    }
   ],
   "source": [
    "cores = 1\n",
    "time = !grep Elapsed ~/metagenome/finished_projects/counting_dbg/kingsford_21/reindeer/nobackup/logs_reruns/*.try4.stderr.log | cut -d' ' -f8\n",
    "mem = !grep 'Maximum resident' /cluster/work/grlab/projects/metagenome/finished_projects/counting_dbg/kingsford_21/reindeer/nobackup/logs_reruns/*.try4.stderr.log | tr -s ' ' | cut -f6 -d' '\n",
    "print(f'BCALM (for REINDEER): {sum([to_seconds(s) for s in time]) / 3600:0.2f} h, in {len(time)} runs, each using {cores} cores\\t'\n",
    "      f'Normalized time: {sum([to_seconds(s) for s in time]) / (36/cores) / 3600:0.3f} h \\tmem: {np.sort([float(m) for m in mem])[::-1][:36//cores].sum() / 1e6} GB')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "e08d010c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-08T09:31:28.410415Z",
     "start_time": "2022-03-08T09:31:27.312138Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(np.array(mem).astype(int) / 1e6)\n",
    "plt.yscale('log')\n",
    "plt.title('BCALM, 1 GB limit')\n",
    "plt.xlabel('RAM usage, GB')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "2be4c3ca",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2022-03-08T11:09:15.458807Z",
     "start_time": "2022-03-08T11:09:14.562628Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/cluster/work/grlab/projects/metagenome/finished_projects/counting_dbg/kingsford_21/reindeer/nobackup/logs_reruns/SRR563547_build_k21.try4.stderr.log'"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ids = !ls /cluster/work/grlab/projects/metagenome/finished_projects/counting_dbg/kingsford_21/reindeer/nobackup/logs_reruns/*.try4.stderr.log\n",
    "ids[np.argmax(np.array(mem).astype(int))]"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.10"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {
    "height": "191px",
    "width": "221px"
   },
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": true,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {
    "height": "calc(100% - 180px)",
    "left": "10px",
    "top": "150px",
    "width": "255px"
   },
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}