Added a realata command

app/reviewphiltrans_app.ml

@@ -4,6 +4,7 @@ open Reviewphiltrans
 let () =
   Command.group ~summary:"Reviewphiltrans" [
     "validation", Pipeline.validation_command ;
+    "realdata", Pipeline.realdata_command ;
     "alistats", Alistats.command ;
   ]
   |> Command.run

lib/convergence_detection.ml

@@ -108,17 +108,17 @@ let plot_merge_results ?t_choices ~plot_all_sites ~(res_by_tools:result list) ~t
   in
   let default_t = List.map res_by_tools ~f:(fun res ->
       let opt = match res with
-        | `Pcoc _ -> "PCOC:1,PC:1,OC:1"
-        | `Pcoc_gamma _ -> "PCOC_gamma:1,PC_gamma:1,OC_gamma:1"
-        | `Pcoc_C60 _ -> "PCOC_C60:1,PC_C60:1,OC_C60:1"
-        | `Diffsel _ -> "Diffsel_mean:0.11,Diffsel_max:1"
-        | `Identical_LG _ -> "Identical_LG08:1"
-        | `Identical_WAG _ -> "Identical_WAG01:1"
-        | `Topological_LG _ -> "Topological_LG08:1"
-        | `Topological_WAG _ -> "Topological_WAG01:1"
-        | `Tdg09 _ -> "Tdg09_1MinusFDR:1,Tdg09_prob_post:1,Tdg09_1MinusLRT:1"
-        | `Multinomial _ -> "Mutinomial_1MinusLRT:1"
-        | `Msd _ -> "Msd_0.05_1MinusP:1"
+        | `Pcoc _ -> "PCOC:0,PC:0,OC:0"
+        | `Pcoc_gamma _ -> "PCOC_gamma:0,PC_gamma:0,OC_gamma:0"
+        | `Pcoc_C60 _ -> "PCOC_C60:0,PC_C60:0,OC_C60:0"
+        | `Diffsel _ -> "Diffsel_mean:0,Diffsel_max:0"
+        | `Identical_LG _ -> "Identical_LG08:0"
+        | `Identical_WAG _ -> "Identical_WAG01:0"
+        | `Topological_LG _ -> "Topological_LG08:0"
+        | `Topological_WAG _ -> "Topological_WAG01:0"
+        | `Tdg09 _ -> "Tdg09_1MinusFDR:0,Tdg09_prob_post:0,Tdg09_1MinusLRT:0"
+        | `Multinomial _ -> "Mutinomial_1MinusLRT:0"
+        | `Msd _ -> "Msd_0.05_1MinusP:0"
 
       in
       string opt

lib/dataset.ml

@@ -16,7 +16,7 @@ let repo ~preview dataset_l =
       let repo_ready_data = Ready_dataset.repo dataset.dataset in
       let repo_raw_data = Raw_dataset.repo ~prefix:model_prefix (Ready_dataset.to_raw dataset.dataset) in
       List.concat [
-        Repo.shift "minimal" (Repo.shift (tree_prefix ^"_"^model_prefix) repo_raw_data);
+        Repo.shift "minimal" (Repo.shift (tree_prefix ^ "_" ^ model_prefix) repo_raw_data);
         Repo.shift "debug" repo_ready_data;
       ]
       |> Repo.shift "dataset"

lib/pipeline.ml

@@ -205,13 +205,13 @@ let derive_from_dataset ~dataset ~preview ~use_diffsel ~use_c60=
       `Pcoc;
     ] @ List.map [0.05] (fun x -> `Msd x)) ;
     (* if preview then *)
-      [`Tdg09;]
+    [`Tdg09;]
     (* else
-      [ `Tdg09;
+       [ `Tdg09;
         `Pcoc_gamma;
         `Identical_WAG;
         `Topological_WAG;
-      ] *)
+       ] *)
     ;
     if use_diffsel then
       [`Diffsel]
@@ -369,3 +369,46 @@ let validation_command =
       in
       validation_main ~outdir ?indir ?ns ?np ?mem ~preview ~use_diffsel ~use_c60 ~no_Ne ~ne_test ~no_HaPC ~tree_dir ~profile_fn ~use_concat ~only_simu ~add_indels ?seed
     ]
+
+let realdata_main ~outdir ~indir ~preview ~use_diffsel ~use_c60 ?(np = 2) ?(mem = 2) ?(seed = Random.int Int.max_value) () =
+  (* seed-related stuff *)
+  printf "Global seed: %i\n" seed;
+  Out_channel.write_all "global.seed" ~data:(string_of_int seed);
+  Random.init seed ;
+
+  (* real trees *)
+  let dataset_l = parse_input_data ~seed indir in
+  let dataset_results_l = derive_det ~dataset_l ~preview ~use_diffsel ~use_c60 in
+  let repo_real_trees = [
+    Dataset.repo dataset_l ~preview ;
+    repo_of_dataset_results_l ~dataset_results_l ;
+  ] |> List.concat (* list of repos *)
+  in
+  let repo = Repo.shift "Real_datasets" repo_real_trees in
+  Repo.build ~outdir ~np ~mem:(`GB mem) ~logger repo
+  (* time_logger#report sim_repo_l#dataset_results (Filename.concat outdir ("elapsed_time_" ^ (Unix.time () |> int_of_float |> string_of_int) ^ ".tsv")) *)
+
+let realdata_command =
+  let open Command.Let_syntax in
+  Command.basic
+    ~summary:"Run simulation pipeline on real data only"
+    [%map_open
+      let outdir =
+        flag "--outdir" (required string) ~doc:"PATH Output directory"
+      and indir =
+        flag "--indir" (required string) ~doc:"PATH Input directory"
+      and preview =
+        flag "--preview-mode" no_arg ~doc:" Preview mode"
+      and use_diffsel =
+        flag "--diffsel" no_arg ~doc:" use the diffsel method (very slow)."
+      and use_c60 =
+        flag "--c60" no_arg ~doc:" use the pcoc c60 method (slow)."
+      and np =
+        flag "--np" (optional int) ~doc:"INT Number of available processors"
+      and mem =
+        flag "--mem" (optional int) ~doc:"INT Available memory (in GB)"
+      and seed =
+        flag "--seed" (optional int) ~doc:"INT Global seed"
+      in
+      realdata_main ~outdir ~indir ?np ?mem ~preview ~use_diffsel ~use_c60 ?seed
+    ]
\ No newline at end of file