bioinfo: added module Star

pveber · Jul 19, 2019 · 6c9397e · 6c9397e
1 parent 0f51f4b
commit 6c9397e
Show file tree

Hide file tree

Showing 2 changed files with 71 additions and 0 deletions.
diff --git a/lib/bioinfo/bistro_bioinfo.ml b/lib/bioinfo/bistro_bioinfo.ml
@@ -2821,3 +2821,59 @@ module Idr = struct
   let items x = Workflow.select x [ "items.tsv" ]
   let figure x = Workflow.select x [ "items.tsv.png" ]
 end
+
+module Star = struct
+  let img = [ docker_image ~account:"flemoine" ~name:"star" () ]
+
+  let mem_in_bytes = seq ~sep:" " [string "$((" ; mem ; string " * 1024 * 1024))$"]
+
+  let genomeGenerate fa =
+    Workflow.shell ~descr:"star.index" ~np:8 ~mem:(Workflow.int (30 * 1024)) [
+      mkdir_p dest ;
+      cmd "STAR" ~img [
+        opt "--runThreadN" Fn.id np ;
+        opt "--runMode" string "genomeGenerate" ;
+        opt "--genomeDir" Fn.id dest ;
+        opt "--genomeFastaFiles" dep fa ;
+        opt "--limitGenomeGenerateRAM" Fn.id mem_in_bytes ;
+      ]
+    ]
+
+  let fq_args = function
+    | `paired_end (fq1, fq2) ->
+      [dep fq1 ; dep fq2]
+    | `single_end fq -> [ dep fq ]
+
+  let samStrandField = function
+    | `None -> string "None"
+    | `intronMotif -> string "intronMotif"
+
+
+  let sorted_mapped_reads x = Workflow.select x ["sorted.bam"]
+
+  let alignReads ?(max_mem = `GB 8)
+      ?outFilterMismatchNmax
+      ?outFilterMultimapNmax
+      ?outSAMstrandField
+      ?alignIntronMax
+      idx fqs =
+    let `GB max_mem = max_mem in
+    Workflow.shell ~descr:"star.map" ~np:8 ~mem:(Workflow.int (max_mem * 1024)) [
+      mkdir_p dest ;
+      cmd "STAR" ~stdout:(dest // "sorted.bam") ~img [
+        opt "--outFileNamePrefix" Fn.id (dest // "star") ;
+        opt "--runThreadN" Fn.id np ;
+        option (opt "--outSAMstrandField" samStrandField) outSAMstrandField ;
+        option (opt "--outFilterMismatchNmax" int) outFilterMismatchNmax ;
+        option (opt "--outFilterMultimapNmax" int) outFilterMultimapNmax ;
+        opt "--genomeDir" dep idx ;
+        opt "--readFilesIn" Fn.id (seq ~sep:" " (fq_args fqs)) ;
+        opt "--outSAMunmapped" string "None" ;
+        opt "--outStd" string "SAM" ;
+        opt "--genomeLoad" string "NoSharedMemory" ;
+        option (opt "--alignIntronMax" int) alignIntronMax ;
+        (* opt "--limitBAMsortRAM" Fn.id mem_in_bytes ; *)
+      ]
+    ]
+    |> sorted_mapped_reads
+end
diff --git a/lib/bioinfo/bistro_bioinfo.mli b/lib/bioinfo/bistro_bioinfo.mli
@@ -916,6 +916,21 @@ module Hisat2 : sig
     sam pworkflow
 end
 
+module Star : sig
+  val genomeGenerate : fasta pworkflow -> [`star_index] dworkflow
+
+  val alignReads :
+    ?max_mem:[`GB of int] ->
+    ?outFilterMismatchNmax:int ->
+    ?outFilterMultimapNmax:int ->
+    ?outSAMstrandField:[`None | `intronMotif] ->
+    ?alignIntronMax:int ->
+    [`star_index] dworkflow ->
+    [ `single_end of sanger_fastq pworkflow
+    | `paired_end of sanger_fastq pworkflow * sanger_fastq pworkflow ] ->
+    bam pworkflow
+end
+
 module Kallisto : sig
   class type index = object
     inherit binary_file