🧹 D3b-ify Sentieon Updates

docs/dockers_ont.md

@@ -12,8 +12,8 @@ nanocaller_merge.cwl|genomicslab/nanocaller:3.2.0
 samtools_coverage.cwl|staphb/samtools:1.17
 samtools_head.cwl|staphb/samtools:1.17
 samtools_split.cwl|staphb/samtools:1.17
-sentieon_LongReadSV.cwl|pgc-images.sbgenomics.com/hdchen/sentieon:202112.06
-sentieon_ReadWriter.cwl|pgc-images.sbgenomics.com/hdchen/sentieon:202112.01_hifi
-sentieon_minimap2.cwl|pgc-images.sbgenomics.com/d3b-bixu/sentieon:202112.01_hifi
+sentieon_DNAscope_LongRead_CLI.cwl|pgc-images.sbgenomics.com/hdchen/sentieon:202308.03
+sentieon_ReadWriter.cwl|pgc-images.sbgenomics.com/hdchen/sentieon:202308.03
+sentieon_minimap2.cwl|pgc-images.sbgenomics.com/hdchen/sentieon:202308.03
 sniffles.cwl|pgc-images.sbgenomics.com/d3b-bixu/sniffles:2.0.7
 tar.cwl|None

docs/dockers_pacbio.md

@@ -10,9 +10,8 @@ pbsv_call.cwl|quay.io/biocontainers/pbsv:2.9.0--h9ee0642_0
 pbsv_discover.cwl|quay.io/biocontainers/pbsv:2.9.0--h9ee0642_0
 samtools_head.cwl|staphb/samtools:1.17
 samtools_split.cwl|staphb/samtools:1.17
-sentieon_DNAscope_LongRead.cwl|pgc-images.sbgenomics.com/hdchen/sentieon:202112.01_hifi
-sentieon_LongReadSV.cwl|pgc-images.sbgenomics.com/hdchen/sentieon:202112.06
-sentieon_ReadWriter.cwl|pgc-images.sbgenomics.com/hdchen/sentieon:202112.01_hifi
-sentieon_minimap2.cwl|pgc-images.sbgenomics.com/d3b-bixu/sentieon:202112.01_hifi
+sentieon_DNAscope_LongRead_CLI.cwl|pgc-images.sbgenomics.com/hdchen/sentieon:202308.03
+sentieon_ReadWriter.cwl|pgc-images.sbgenomics.com/hdchen/sentieon:202308.03
+sentieon_minimap2.cwl|pgc-images.sbgenomics.com/hdchen/sentieon:202308.03
 sniffles.cwl|pgc-images.sbgenomics.com/d3b-bixu/sniffles:2.0.7
 tar.cwl|None

scripts/get_dnascope_model.py

@@ -0,0 +1,28 @@
+#!/usr/bin/env python3
+
+import argparse
+import yaml
+import requests
+import sys
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Download DNAscope model bundle")
+    parser.add_argument("model_name", help="the name of the model bundle, e.g. Illumina_WGS")
+    args = parser.parse_args()
+    model_name = args.model_name.split("-")
+    sentieon_models_yaml = "https://github.com/Sentieon/sentieon-models/raw/refs/heads/main/sentieon_models.yaml"
+    response = requests.get(sentieon_models_yaml, allow_redirects=True)
+    content = response.content.decode("utf-8")
+    content = yaml.safe_load(content)
+    try:
+        url = content["DNAscope_bundles"][model_name[0]][model_name[1]]
+        r = requests.get(url, allow_redirects=True)
+        open(url.split("/")[-1], 'wb').write(r.content)
+    except:
+        open('empty.bundle', 'wb')
+    print('Models updated on: ' + content["Updated on"], file=sys.stderr)
+
+
+if __name__ == '__main__':
+    main()

tools/download_DNAscope_model.cwl

@@ -0,0 +1,46 @@
+cwlVersion: v1.2
+class: CommandLineTool
+label: Download DNAscope model bundle
+hints:
+  - class: ResourceRequirement
+    coresMin: 1
+requirements:
+  - class: ShellCommandRequirement
+  - class: InlineJavascriptRequirement
+  - class: DockerRequirement
+    dockerPull: python:3.7-slim
+  - class: InitialWorkDirRequirement
+    listing:
+      - entryname: get_dnascope_model.py
+        entry:
+          $include: ../scripts/get_dnascope_model.py
+arguments:
+  - position: 0
+    valueFrom: 'pip install pyyaml requests;'
+    shellQuote: false
+  - position: 1
+    valueFrom: 'python get_dnascope_model.py'
+    shellQuote: false
+inputs:
+  - id: model_name
+    label: Model name
+    doc: Model platform and data type. For example, Illumina_WGS
+    type: 
+    - type: enum
+      symbols:
+        - Illumina-WGS
+        - Illumina-WES
+        - MGI-WGS
+        - MGI-WES
+        - Element_Biosciences-WGS
+        - PacBio_HiFi-WGS
+        - Oxford_Nanopore-WGS
+    inputBinding:
+      position: 2
+outputs:
+  - id: model_bundle
+    label: DNAscope Model bundle
+    type: File
+    outputBinding:
+      glob: '*.bundle'
+