bioinfo: introduced SE_or_PE type

doc/getting-started.rst

@@ -77,7 +77,7 @@ paste the following program:
    let genome = Ucsc_gb.genome_sequence `sacCer2    (* Fetch a reference genome *)
    let bowtie2_index = Bowtie2.bowtie2_build genome (* Build a Bowtie2 index from it *)
    let sample_sam =                                 (* Map the reads on the reference genome *)
-     Bowtie2.bowtie2 bowtie2_index (`single_end [ sample_fq ])
+     Bowtie2.bowtie2 bowtie2_index (SE_or_PE.Single_end [ sample_fq ])
    let sample_peaks =                               (* Call peaks on mapped reads *)
      Macs2.(callpeak sam [ sample_sam ])
 

examples/chipseq.ml

@@ -1,4 +1,4 @@
-(** 
+(**
    Paper: https://www.ncbi.nlm.nih.gov/pubmed/21700227
    Datasets: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE29506
 *)
@@ -13,7 +13,7 @@ let bowtie_index = Bowtie.bowtie_build genome
 
 let mapped_reads srrid =
   let fastq = Sra_toolkit.fastq_dump (`id srrid) in
-  Bowtie.bowtie ~v:1 bowtie_index (`single_end [ fastq ])
+  Bowtie.bowtie ~v:1 bowtie_index (SE_or_PE.Single_end [ fastq ])
 
 let macs2 =
   Macs2.callpeak ~qvalue:1e-10 ~mfold:(1,100) Macs2.sam
@@ -44,5 +44,5 @@ let () =
     ~allowed_containers:[`Singularity]
     ~np:4 ~mem:(`GB 4)
     ~outdir:"res"
-    ~loggers:[ Console_logger.create () ] 
+    ~loggers:[ Console_logger.create () ]
     repo

examples/rnaseq.ml

@@ -39,7 +39,7 @@ let genome : fasta pworkflow =
 let bowtie2_index = Bowtie2.bowtie2_build genome
 
 let mapped_reads x =
-  Bowtie2.bowtie2 bowtie2_index (`single_end [ fastq x ])
+  Bowtie2.bowtie2 bowtie2_index (SE_or_PE.Single_end [ fastq x ])
 
 let annotation : gff pworkflow =
   Bistro_unix.wget "ftp://ftp.ensemblgenomes.org/pub/bacteria/release-41/gff3/bacteria_21_collection/listeria_monocytogenes_10403s/Listeria_monocytogenes_10403s.ASM16869v2.41.chromosome.Chromosome.gff3.gz"

examples/zhou2011.ml

@@ -35,7 +35,7 @@ let fastq x = List.map (srr_id x) ~f:(fun id ->
 let bowtie_index = Bowtie.bowtie_build genome
 
 let mapped_reads x =
-  Bowtie.bowtie ~v:1 bowtie_index (`single_end (fastq x))
+  Bowtie.bowtie ~v:1 bowtie_index (SE_or_PE.Single_end (fastq x))
 
 let mapped_reads_bam x =
   Samtools.indexed_bam_of_sam (mapped_reads x)

lib/bioinfo/bistro_bioinfo.ml

@@ -116,6 +116,16 @@ module Bed = struct
   let keep6 x = keep ~n:6 x
 end
 
+module SE_or_PE = struct
+  type 'a t =
+    | Single_end of 'a
+    | Paired_end of 'a * 'a
+
+  let map x ~f = match x with
+    | Single_end x -> Single_end (f x)
+    | Paired_end (x, y) -> Paired_end (f x, f y)
+end
+
 module Fastq = struct
 
   type _ format =
@@ -402,9 +412,9 @@ module Bowtie2 = struct
       ?fastq_format index fqs =
 
     let args = match fqs with
-      | `single_end fqs ->
+      | SE_or_PE.Single_end fqs ->
         opt "-U" (list dep ~sep:",") fqs
-      | `paired_end (fqs1, fqs2) ->
+      | Paired_end (fqs1, fqs2) ->
         seq [
           opt "-1" (list dep ~sep:",") fqs1 ;
           string " " ;
@@ -474,8 +484,8 @@ module Bowtie = struct
 
   let bowtie ?l ?e ?m ?fastq_format ?n ?v ?maxins index fastq_files =
     let args = match fastq_files with
-      | `single_end fqs -> list dep ~sep:"," fqs
-      | `paired_end (fqs1, fqs2) ->
+      | SE_or_PE.Single_end fqs -> list dep ~sep:"," fqs
+      | Paired_end (fqs1, fqs2) ->
         seq [
           opt "-1" (list dep ~sep:",") fqs1 ;
           string " " ;
@@ -1963,8 +1973,8 @@ module Tophat = struct
 
   let tophat1 ?color index fqs =
     let args = match fqs with
-      | `single_end fqs -> list dep ~sep:"," fqs
-      | `paired_end (fqs1, fqs2) ->
+      | SE_or_PE.Single_end fqs -> list dep ~sep:"," fqs
+      | Paired_end (fqs1, fqs2) ->
         seq [
           list dep ~sep:"," fqs1 ;
           string " " ;
@@ -1984,8 +1994,8 @@ module Tophat = struct
 
   let tophat2 index fqs =
     let args = match fqs with
-      | `single_end fqs -> list dep ~sep:"," fqs
-      | `paired_end (fqs1, fqs2) ->
+      | SE_or_PE.Single_end fqs -> list dep ~sep:"," fqs
+      | Paired_end (fqs1, fqs2) ->
         seq [
           list dep ~sep:"," fqs1 ;
           string " " ;
@@ -2695,9 +2705,9 @@ module Hisat2 = struct
       fqs
     =
     let args = match fqs with
-      | `single_end fqs ->
+      | SE_or_PE.Single_end fqs ->
         opt "-U" (list dep ~sep:",") fqs
-      | `paired_end (fqs1, fqs2) ->
+      | Paired_end (fqs1, fqs2) ->
         seq [
           opt "-1" (list dep ~sep:",") fqs1 ;
           string " " ;
@@ -2838,9 +2848,9 @@ module Star = struct
     ]
 
   let fq_args = function
-    | `paired_end (fq1, fq2) ->
+    | SE_or_PE.Single_end fq -> [ dep fq ]
+    | Paired_end (fq1, fq2) ->
       [dep fq1 ; dep fq2]
-    | `single_end fq -> [ dep fq ]
 
   let samStrandField = function
     | `None -> string "None"

lib/bioinfo/bistro_bioinfo.mli

@@ -92,6 +92,14 @@ class type sra = object
   method format : [`sra]
 end
 
+module SE_or_PE : sig
+  type 'a t =
+    | Single_end of 'a
+    | Paired_end of 'a * 'a
+
+  val map : 'a t -> f:('a -> 'b) -> 'b t
+end
+
 (** {3 File_formats} *)
 
 module Bed : sig
@@ -792,8 +800,7 @@ module Bowtie : sig
     ?n:int -> ?v:int ->
     ?maxins:int ->
     index pworkflow ->
-    [ `single_end of 'a pworkflow list
-    | `paired_end of 'a pworkflow list * 'a pworkflow list ] ->
+    'a pworkflow list SE_or_PE.t ->
     sam pworkflow
 end
 
@@ -847,8 +854,7 @@ module Bowtie2 : sig
     ?seed:int ->
     ?fastq_format:'a Fastq.format ->
     index pworkflow ->
-    [ `single_end of 'a pworkflow list
-    | `paired_end of 'a pworkflow list * 'a pworkflow list ] ->
+    'a pworkflow list SE_or_PE.t ->
     sam pworkflow
 end
 
@@ -862,14 +868,12 @@ module Tophat : sig
   val tophat1 :
     ?color:bool ->
     Bowtie.index pworkflow ->
-    [ `single_end of #fastq pworkflow list
-    | `paired_end of (#fastq as 'a) pworkflow list * 'a pworkflow list ] ->
+    #fastq pworkflow list SE_or_PE.t ->
     output pworkflow
 
   val tophat2 :
     Bowtie2.index pworkflow ->
-    [ `single_end of #fastq pworkflow list
-    | `paired_end of (#fastq as 'a) pworkflow list * 'a pworkflow list ] ->
+    #fastq pworkflow list SE_or_PE.t ->
     output pworkflow
 
   val accepted_hits : output pworkflow -> bam pworkflow
@@ -911,8 +915,7 @@ module Hisat2 : sig
     ?no_discordant:bool ->
     ?seed:int ->
     [`hisat2_index] dworkflow ->
-    [ `single_end of sanger_fastq pworkflow list
-    | `paired_end of sanger_fastq pworkflow list * sanger_fastq pworkflow list ] ->
+    sanger_fastq pworkflow list SE_or_PE.t ->
     sam pworkflow
 end
 
@@ -926,8 +929,7 @@ module Star : sig
     ?outSAMstrandField:[`None | `intronMotif] ->
     ?alignIntronMax:int ->
     [`star_index] dworkflow ->
-    [ `single_end of sanger_fastq pworkflow
-    | `paired_end of sanger_fastq pworkflow * sanger_fastq pworkflow ] ->
+    sanger_fastq pworkflow SE_or_PE.t ->
     bam pworkflow
 end