bioinfo: added Idr

lib/bioinfo/bistro_bioinfo.ml

@@ -2761,3 +2761,63 @@ module Picardtools = struct
       ]
     ]
 end
+
+module Idr = struct
+  type 'a format = NarrowPeak | BroadPeak | Bed | Gff
+
+  let narrowPeak = NarrowPeak
+  let broadPeak = BroadPeak
+  let bed = Bed
+  let gff = Gff
+
+  type 'a output = [`idr_output of 'a]
+
+  let string_of_file_format = function
+    | NarrowPeak -> "narrowPeak"
+    | BroadPeak -> "broadPeak"
+    | Bed -> "bed"
+    | Gff -> "gff"
+
+  let file_format x = string (string_of_file_format x)
+
+  let string_of_merge_method = function
+    | `sum -> "sum"
+    | `avg -> "avg"
+    | `min -> "min"
+    | `max -> "max"
+
+  let merge_method x = string (string_of_merge_method x)
+
+  let token_of_rank r =
+    string (
+      match r with
+      | `signal -> "signal.value"
+      | `pvalue -> "p.value"
+      | `qvalue -> "q.value"
+    )
+
+  let img = [ docker_image ~account:"pveber" ~name:"idr" ~tag:"2.0.3" () ]
+
+  let idr
+      ~input_file_type ?idr_threshold ?soft_idr_threshold
+      ?peak_merge_method ?rank ?random_seed ?peak_list
+      sample1 sample2 =
+    Workflow.shell ~descr:"Idr.idr" [
+      mkdir_p dest ;
+      cmd "idr" ~img [
+        opt "--input-file-type" file_format input_file_type ;
+        opt "--output-file" (fun x -> x) (dest // "items.tsv") ;
+        option (opt "--idr-threshold" float) idr_threshold ;
+        option (opt "--soft-idr-threshold" float) soft_idr_threshold ;
+        option (opt "--peak-merge-method" merge_method) peak_merge_method ;
+        option (opt "--rank" token_of_rank) rank ;
+        option (opt "--random-seed" int) random_seed ;
+        option (opt "--peak-list" dep) peak_list ;
+        string "--plot" ;
+        opt "--samples" (list ~sep:" " dep) [ sample1 ; sample2 ] ;
+      ]
+    ]
+
+  let items x = Workflow.select x [ "items.tsv" ]
+  let figure x = Workflow.select x [ "items.tsv.png" ]
+end

lib/bioinfo/bistro_bioinfo.mli

@@ -1215,6 +1215,31 @@ module Macs2 : sig
   val broad_peaks : broad_output pworkflow -> broad_peaks pworkflow
 end
 
+module Idr : sig
+  type 'a format
+
+  val narrowPeak : Macs2.narrow_peaks format
+  val broadPeak : Macs2.broad_peaks format
+  val bed : bed3 format
+  val gff : gff format
+
+  type 'a output = [`idr_output of 'a]
+
+  val idr :
+    input_file_type:'a format ->
+    ?idr_threshold:float ->
+    ?soft_idr_threshold:float ->
+    ?peak_merge_method:[ `sum | `avg | `min | `max] ->
+    ?rank:[ `signal | `pvalue | `qvalue ] ->
+    ?random_seed:int ->
+    ?peak_list:'a pworkflow ->
+    'a pworkflow ->
+    'a pworkflow ->
+    'a output dworkflow
+
+  val items : 'a output dworkflow -> 'a pworkflow
+  val figure : _ output dworkflow -> png pworkflow
+end
 
 module Meme_suite : sig
   class type meme_output = object