Merge PR #106 (@ihodes/@smondet, mouse genomes)

This closes #103 also.

src/lib/build_machine.ml

@@ -43,6 +43,8 @@ let create
           ~host ~run_program ~destination_path:(meta_playground // "hg19-reference-genome")
       | `B37decoy -> Data_providers.pull_b37decoy
                        ~host ~run_program ~destination_path:(meta_playground // "hs37d5-reference-genome")
+      | `mm10 -> Data_providers.pull_mm10
+                   ~host ~run_program ~destination_path:(meta_playground // "mm10-reference-genome")
       )
     ~host
     ~toolkit:(

src/lib/download_reference_genomes.ml

@@ -216,3 +216,12 @@ let pull_hg18 ~host ~(run_program : Machine.run_function) ~destination_path =
     wget_gunzip ~host ~run_program dbsnp_hg18_url
       ~destination:(destination_path // "dbsnp.vcf") in
   Reference_genome.create "hg18" fasta ~dbsnp
+
+let mm10_url =
+  "ftp://ftp.ncbi.nlm.nih.gov/genomes/all/GCA_000001635.6_GRCm38.p4/GCA_000001635.6_GRCm38.p4_genomic.fna.gz"
+
+let pull_mm10 ~host ~(run_program : Machine.run_function) ~destination_path =
+  let fasta =
+    wget_gunzip ~host ~run_program hg18_url
+      ~destination:(destination_path // "mm10.fasta") in
+  Reference_genome.create "mm10" fasta

src/lib/download_reference_genomes.mli

@@ -24,3 +24,8 @@ val pull_hg19 :
   host:Common.KEDSL.Host.t ->
   run_program:Run_environment.Machine.run_function ->
   destination_path:string -> Reference_genome.t
+
+val pull_mm10 :
+  host:Common.KEDSL.Host.t ->
+  run_program:Run_environment.Machine.run_function ->
+  destination_path:string -> Reference_genome.t

src/lib/reference_genome.ml

@@ -1,7 +1,7 @@
 open Common
 
 type specification =
-  [`B37 | `B38 | `hg19 | `hg18 | `B37decoy ]
+  [`B37 | `B38 | `hg19 | `hg18 | `B37decoy | `mm10 ]
 
 (** A reference genome has a name (for display/matching) and a
      cluster-dependent path.
@@ -36,12 +36,12 @@ let on_host ~host ?cosmic ?dbsnp ?gtf ?cdna name path =
 
 let name t = t.name
 let path t = t.location#product#path
-let cosmic_path_exn t = 
+let cosmic_path_exn t =
   let msg = sprintf "cosmic_path_exn of %s" t.name in
   let cosmic = Option.value_exn ~msg t.cosmic in
   cosmic#product#path
 
-let dbsnp_path_exn t = 
+let dbsnp_path_exn t =
   let msg = sprintf "dbsnp_path_exn of %s" t.name in
   let trgt = Option.value_exn ~msg t.dbsnp in
   trgt#product#path

src/lib/reference_genome.mli

@@ -21,7 +21,7 @@
 open Common
 
 type specification =
-  [`B37 | `B38 | `hg19 | `hg18 | `B37decoy ]
+  [`B37 | `B38 | `hg19 | `hg18 | `B37decoy | `mm10 ]
 
 type t = private {
   name : string;
@@ -47,8 +47,8 @@ val create :
 
 val on_host :
   host:KEDSL.Host.t ->
-  ?cosmic:string -> 
-  ?dbsnp:string -> 
+  ?cosmic:string ->
+  ?dbsnp:string ->
   ?gtf:string->
   ?cdna:string->
   string -> string -> t
@@ -63,7 +63,7 @@ val cosmic_path_exn : t -> string
 val dbsnp_path_exn : t -> string
 val gtf_path_exn : t -> string
 val cdna_path_exn : t -> string
- 
+
 (** {5 Targets} *)
 
 val fasta: t -> KEDSL.file_workflow

src/lib/region.ml

@@ -40,32 +40,32 @@ let to_samtools_option r =
   | Some s -> sprintf "-r %s" s
   | None -> ""
 
-let to_gatk_option r = 
+let to_gatk_option r =
   match to_samtools_specification r with
   | Some s -> sprintf "--intervals %s" s
   | None -> ""
 
 let parse_samtools s =
   match String.split ~on:(`Character ':') s with
   | [] -> assert false
-  | [one] -> `Chromosome one 
-  | [one; two] -> 
+  | [one] -> `Chromosome one
+  | [one; two] ->
     begin match String.split ~on:(`Character '-') two with
-    | [left; right] -> 
+    | [left; right] ->
       begin match Int.of_string left, Int.of_string right with
       | Some b, Some e -> `Chromosome_interval (one, b, e)
       | _ -> failwithf "Cannot parse %S into 2 loci" two
       end
     | _ -> failwithf "Not one '-' in %S" two
-    end 
+    end
   | _ -> failwithf "Not one or zero ':' in %S" s
 
 
 let cmdliner_term () =
   let open Cmdliner in
   Term.(
     pure (function
-      | None -> `Full 
+      | None -> `Full
       | Some s -> parse_samtools s)
     $ Arg.(
         value & opt (some string) None
@@ -128,13 +128,35 @@ let all_chromosomes_hg19 = [
   `Chromosome "chrX";
   `Chromosome "chrY";
   `Chromosome "chrM";
+  ]
+
+let all_chromosomes_mm10 = [
+  `Chromosome "1";
+  `Chromosome "2";
+  `Chromosome "3";
+  `Chromosome "4";
+  `Chromosome "5";
+  `Chromosome "6";
+  `Chromosome "7";
+  `Chromosome "8";
+  `Chromosome "9";
+  `Chromosome "10";
+  `Chromosome "11";
+  `Chromosome "12";
+  `Chromosome "13";
+  `Chromosome "14";
+  `Chromosome "15";
+  `Chromosome "16";
+  `Chromosome "17";
+  `Chromosome "18";
+  `Chromosome "19";
+  `Chromosome "X";
+  `Chromosome "Y";
 ]
 
-let major_contigs ~reference_build = 
-  match reference_build with 
+let major_contigs ~reference_build =
+  match reference_build with
     | (`B37 | `B37decoy) -> all_chromosomes_b37
     | `B38 -> all_chromosomes_b37
     | (`hg19 | `hg18) -> all_chromosomes_hg19
-
-
-
+    | `mm10 -> all_chromosomes_mm10

src/lib/run_environment.ml

@@ -122,6 +122,3 @@ module Machine = struct
   let work_dir t = t.work_dir
 
 end
-
-
-