start the results merging step

lib/convergence_detection.ml

@@ -3,6 +3,7 @@ open Bistro.EDSL
 open Bistro.Std
 open File_formats
 open Bistro_bioinfo.Std
+open Defs
 
 type pcoc_out
 type diffsel_out
@@ -11,6 +12,19 @@ type det_out =
   | Pcoc_out
   | Diffsel_out
 
+type det_result = {
+  dataset : dataset ;
+  det_meth : det_meth ;
+  det_result : det_out directory workflow ;
+  }
+
+type dataset_res = {
+  model_prefix : string ;
+  tree_prefix : string ;
+  res_by_tools: det_result list ;
+  merged_results : text_file workflow
+  }
+
 let pcoc ?plot_complete ?gamma  ~(faa:aminoacid_fasta workflow) ~(tree:_ workflow) :  (*`pcoc TODO*) det_out directory workflow =
   let env = docker_image ~account:"carinerey" ~name:"pcoc" ~tag:"06212018" () in
   let run_pcoc = workflow ~descr:"convergence_detection.pcoc" [
@@ -69,3 +83,30 @@ let diffsel ~(phy_n:nucleotide_phylip workflow) ~(tree: _ workflow) ~(w_every:in
       ]
     )
   ]
+
+
+let merge_results ~res_by_tools : text_file workflow =
+  let command = List.map res_by_tools ~f:(fun res ->
+    let def_meth = res.det_meth in
+    let w = match def_meth with
+      | Pcoc -> res.det_result
+      | Pcoc_gamma -> res.det_result
+      | Diffsel -> res.det_result
+    in
+    let opt = match def_meth with
+      | Pcoc -> string "--pcoc"
+      | Pcoc_gamma -> string "--pcoc_gamma"
+      | Diffsel -> string "--diffsel"
+    in
+    seq ~sep:" " [opt; dep w]
+  )
+  in
+  workflow ~descr:"convergence_detection.merge_results" [
+    cmd "python"  [
+            file_dump (string Scripts.merge_det_results) ;
+            opt "-o" ident dest ;
+            seq ~sep:" " command ;
+          ] ;
+  ]
+
+

lib/convergence_detection.mli

@@ -3,6 +3,7 @@ open Bistro.EDSL
 open Bistro.Std
 open Bistro_bioinfo.Std
 open File_formats
+open Defs
 
 type pcoc_out
 type diffsel_out
@@ -11,6 +12,19 @@ type det_out =
   | Pcoc_out
   | Diffsel_out
 
+type det_result = {
+  dataset : dataset ;
+  det_meth : det_meth ;
+  det_result : det_out directory workflow ;
+  }
+
+type dataset_res = {
+  model_prefix : string ;
+  tree_prefix : string ;
+  res_by_tools: det_result list ;
+  merged_results : text_file workflow
+  }
+
 val pcoc :
   ?plot_complete : bool ->
   ?gamma : bool ->
@@ -24,3 +38,8 @@ val diffsel :
   w_every : int ->
   n_cycles: int ->
   (*[`diffsel]*) det_out directory workflow
+
+
+val merge_results :
+  res_by_tools : det_result list ->
+  text_file workflow

lib/defs.ml

@@ -33,7 +33,6 @@ type dataset = {
   ready_dataset : ready_dataset
   }
 
-
 type det_meth =
   | Pcoc
   | Pcoc_gamma
@@ -49,14 +48,5 @@ type 'a w_det_meth =
   | Pcoc_gamma_w of 'a
   | Diffsel of 'a
 
-type det_result = {
-  dataset : dataset ;
-  det_meth : det_meth ;
-  det_result : Convergence_detection.det_out directory workflow ;
-  }
 
-type dataset_res = {
-  res_by_tools: det_result list ;
-  merged_results : string
-  }
 

lib/pipeline.ml

@@ -4,6 +4,7 @@ open Bistro.EDSL
 open Bistro.Std
 open File_formats
 open Defs
+open Convergence_detection
 
 let ready_dataset_of_raw_dataset raw_dataset =
     let input_tree = raw_dataset.input_tree in
@@ -39,7 +40,7 @@ let repo_of_raw_dataset (raw_dataset:raw_dataset) =
 
 
 let repo_of_dataset_l ~preview dataset_l =
-  List.map dataset_l ~f:(fun dataset ->
+  List.map dataset_l ~f:(fun (dataset : Defs.dataset) ->
     let model_prefix = dataset.model_prefix in
     let tree_prefix = dataset.tree_prefix in
     let w_dataset =
@@ -91,12 +92,15 @@ let derive_sim ~tree_dir ~trees ~profile_fn ~preview =
 let repo_of_dataset_results_l ~dataset_results_l =
   List.map dataset_results_l ~f:(fun dataset_results ->
     let det_results_l = dataset_results.res_by_tools in
+    let merged_results = dataset_results.merged_results in
+    let merged_results_item = Repo.shift dataset_results.tree_prefix (Repo.shift dataset_results.model_prefix [(Repo.item ["merged_results.tsv"] merged_results) ]) in
+    [ merged_results_item ;
     List.map det_results_l ~f:(fun det_results ->
       let model_prefix = det_results.dataset.model_prefix in
       let tree_prefix = det_results.dataset.tree_prefix in
       let det_meth = det_results.det_meth in
       let det_meth_prefix = string_of_det_meth det_meth in
-      let open Convergence_detection in
+
       let w = det_results.det_result
         in
       let repo_d = Repo.shift "Detection_tools" (Repo.shift det_meth_prefix (
@@ -110,11 +114,11 @@ let repo_of_dataset_results_l ~dataset_results_l =
       in
       Repo.shift tree_prefix (Repo.shift model_prefix repo_d)
     ) |> List.concat
+    ] |> List.concat
   )
   |> List.concat
 
 let derive_from_det_meth ~det_meth ~dataset ~preview =
-  let open Convergence_detection in
   let faa = dataset.ready_dataset.faa in
   let fna = dataset.ready_dataset.fna in
   let phy_n = Bppsuite.fa2phy ~fna in
@@ -132,13 +136,15 @@ let derive_from_det_meth ~det_meth ~dataset ~preview =
 
 
 
-let derive_from_dataset ~dataset ~preview=
+let derive_from_dataset ~dataset ~preview =
   let det_meths = [Pcoc;Pcoc_gamma;Diffsel] in
   let res_by_tools = List.map det_meths ~f:(fun det_meth ->
     derive_from_det_meth ~det_meth ~dataset ~preview
     ) in
-  let merged_results = "" (* TODO: replace "" by a workflow *) in
-  {res_by_tools; merged_results}
+  let merged_results = merge_results ~res_by_tools in
+  let model_prefix = dataset.model_prefix in
+  let tree_prefix = dataset.tree_prefix in
+  {model_prefix; tree_prefix; res_by_tools ; merged_results}
 
 let derive_det ~dataset_l ~profile_fn ~preview=
   List.map dataset_l ~f:(fun dataset ->