add a post analyse : simu_infos

lib/convergence_detection.ml

@@ -33,6 +33,7 @@ let meth_string_of_result = function
 type dataset_res = {
   model_prefix : string ;
   tree_prefix : string ;
+  dataset : Dataset.t ;
   res_by_tools: result list ;
   merged_results : text_file workflow ;
   plot_merged_results : svg workflow ;

lib/convergence_detection.mli

@@ -5,7 +5,6 @@ open Bistro_bioinfo.Std
 open File_formats
 open Defs
 
-
 type result = [
   | `Pcoc of [`pcoc] directory workflow
   | `Pcoc_gamma of [`pcoc] directory workflow
@@ -24,6 +23,7 @@ val meth_string_of_result : result -> string
 type dataset_res = {
   model_prefix : string ;
   tree_prefix : string ;
+  dataset : Dataset.t ;
   res_by_tools: result list ;
   merged_results : text_file workflow ;
   plot_merged_results : svg workflow

lib/dataset.ml

@@ -15,8 +15,8 @@ let repo ~preview dataset_l =
       let repo_ready_data = Ready_dataset.repo dataset.dataset in
       let repo_raw_data = if preview then Raw_dataset.repo ~prefix:model_prefix (Ready_dataset.to_raw dataset.dataset) else [] in
         List.concat [
-          Repo.shift "data" (Repo.shift (tree_prefix ^"_"^model_prefix) repo_raw_data);
-          Repo.shift "debug_data" (Repo.shift tree_prefix (Repo.shift model_prefix repo_ready_data));
+          Repo.shift "simulated_data" (Repo.shift (tree_prefix ^"_"^model_prefix) repo_raw_data);
+          Repo.shift "simulated_data_debug" (Repo.shift tree_prefix (Repo.shift model_prefix repo_ready_data));
         ]
     )
   |> List.concat

lib/pipeline.ml

@@ -217,7 +217,7 @@ let derive_from_dataset ~dataset ~preview ~no_diffsel=
   let plot_merged_results = plot_merge_results ~plot_all_sites ~res_by_tools ~tsv ~faa ~tree in
   let model_prefix = dataset.model_prefix in
   let tree_prefix = dataset.tree_prefix in
-  {model_prefix; tree_prefix; res_by_tools ; merged_results ; plot_merged_results}
+  {model_prefix; tree_prefix; dataset; res_by_tools ; merged_results ; plot_merged_results}
 
 let derive_det ~dataset_l ~preview ~no_diffsel =
   List.map dataset_l ~f:(fun dataset ->

lib/post_analyses.ml

@@ -8,15 +8,25 @@ open Convergence_detection
 
 type post_analyses_dir
 
-type post_analyses = {
+type t_choices = {
   t_choices_complete: text_file workflow ;
   t_choices_max: text_file workflow ;
   t_choices_plot: text_file workflow ;
   }
   
-let is_hyp ~hyp dataset_results = 
+type simu_infos = {
+  simu_infos: text_file workflow option ;
+  model_prefix: string ;
+  }
+
+type post_analyses = {
+  t_choices : t_choices option;
+  simu_infos_l : simu_infos list;
+  }
+
+
+let is_hyp ~hyp (dataset_results :dataset_res) =
   let model_prefix = dataset_results.model_prefix in
-  let merged_results = dataset_results.merged_results in
   model_prefix = hyp
 
 let make_t_choices ~h0_merged_results ~ha_merged_results : post_analyses_dir directory workflow =
@@ -35,7 +45,19 @@ let make_t_choices ~h0_merged_results ~ha_merged_results : post_analyses_dir dir
     ])
   ]
 
-let post_analyses_of_dataset_results_l ~dataset_results_l =
+let make_simu_infos ?(descr="") ?(fna_infos) ~faa ~tree_sc : text_file workflow =
+  let env = docker_image ~account:"carinerey" ~name:"pcoc" ~tag:"07022018" () in
+  workflow ~descr:("post_analyses.simu_infos." ^ descr) [
+    cmd "python" ~env [
+          file_dump (string Scripts.calc_simu_infos) ;
+          opt "--faa" dep faa;
+          opt "--tree" dep tree_sc;
+          option ( opt "--fna_infos" dep) fna_infos;
+          opt "--output " ident dest;
+    ];
+  ]
+
+let get_t_choices ~(dataset_results_l: dataset_res list) : t_choices option =
   let h0_res = List.find dataset_results_l (is_hyp ~hyp: "H0") in
   let ha_res = List.find dataset_results_l (is_hyp ~hyp: "HaPCOC") in
   match (h0_res, ha_res) with
@@ -47,14 +69,42 @@ let post_analyses_of_dataset_results_l ~dataset_results_l =
       let t_choices_complete = t_choices_dir / selector ["out.complete.tsv"] in
       let t_choices_plot = t_choices_dir / selector ["out.pdf"] in
       Some {t_choices_max; t_choices_complete ; t_choices_plot}
-   | _ -> failwith {| tata |}
+   | _ -> None
+
+let get_simu_infos ~dataset_results =
+  let model_prefix = dataset_results.dataset.model_prefix in
+  let ready_dataset = dataset_results.dataset.dataset in
+  let faa = ready_dataset.faa in
+  let tree_sc = Tree_dataset.tree ready_dataset.tree_dataset `Detection in
+  let fna_infos = ready_dataset.fna_infos in
+  match fna_infos with
+   | Some w -> Some (make_simu_infos ~descr:model_prefix ~faa ~tree_sc ~fna_infos:w)
+   | None -> None (*make_simu_infos ~faa ~tree_sc*)
+
+
+let post_analyses_of_dataset_results_l ~dataset_results_l =
+  let t_choices = get_t_choices ~dataset_results_l in
+  let simu_infos_l = List.map dataset_results_l ~f:(fun dataset_results ->
+  {simu_infos = (get_simu_infos ~dataset_results); model_prefix = dataset_results.model_prefix} )in
+  {t_choices ; simu_infos_l}
 
 let repo_of_post_analyses ~prefix ~post_analyses =
-  match post_analyses with
+  [
+  (match post_analyses.t_choices with
     | None -> []
-    | Some w -> 
+    | Some w ->
           Repo.[
             item [prefix ^ ".t_choices.max_mcc_per_meth.tsv"] w.t_choices_max ;
             item [prefix ^ ".t_choices.complete.tsv"] w.t_choices_complete ;
             item [prefix ^ ".t_choices.pdf"] w.t_choices_plot ;
           ] |> Repo.shift "t_choices"
+  );
+  List.map post_analyses.simu_infos_l ~f:(fun simu_infos ->
+    match simu_infos.simu_infos with
+    | None -> []
+    | Some w ->
+          Repo.[
+            item [prefix ^ "." ^ simu_infos.model_prefix ^ ".tsv"] w
+          ] |> Repo.shift "simu_infos"
+    ) |> List.concat
+  ] |> List.concat