refactoring

lib/convergence_detection.ml

@@ -28,7 +28,7 @@ type dataset_res = {
 
 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" [
+  workflow ~descr:"convergence_detection.pcoc" [
     cmd "pcoc_det.py" ~env [
         opt "-t" dep tree;
         opt "-m" string "-";
@@ -37,7 +37,10 @@ let pcoc ?plot_complete ?gamma  ~(faa:aminoacid_fasta workflow) ~(tree:_ workflo
         option ( flag string "--gamma" ) gamma;
         option ( flag string "--plot --plot_complete_ali" ) plot_complete;
     ]
-  ] (* TODO / selector ["RUN*/*.results.tsv"] *) in
+  ]
+
+
+let pcoc_selector (run_pcoc:det_out directory workflow) : text_file workflow =
   let out_pcoc = (dep run_pcoc) // "RUN*/*.results.tsv" in
   workflow ~descr:"convergence_detection.selector_pcoc" [
     cmd "cp" [out_pcoc; ident dest] ;
@@ -46,20 +49,39 @@ let pcoc ?plot_complete ?gamma  ~(faa:aminoacid_fasta workflow) ~(tree:_ workflo
 
 let diffsel ~(phy_n:nucleotide_phylip workflow) ~(tree: _ workflow) ~(w_every:int) ~(n_cycles: int) :  (*`diffsel TODO*) det_out directory workflow =
   let env = docker_image ~account:"vlanore" ~name:"diffsel" ~tag:"v1.0" () in
+  let tmp_tree = tmp // "myrun.tree" in
+  let tmp_ali = tmp // "myrun.ali" in
+  let dest_tree = dest // "myrun.tree" in
+  let dest_ali = dest // "myrun.ali" in
   let chainname = dest // "myrun" in
   (*_build/diffsel -t data/samhd1.tree -d data/samhd1.ali -ncond 3 -x 1 10000 myrun*)
-  let run_diffsel = workflow ~descr:"convergence_detection.run_diffsel" [
-    mkdir_p dest;
-    cmd "_build/diffsel" ~env [
-        opt "-t" dep tree;
-        opt "-d"  dep phy_n ;
-        opt "-ncond"  int 2 ;
-        opt "-x" seq [ int w_every; string " "; int n_cycles];
-        ident chainname ;
-    ];
-  ] in
+  workflow ~descr:"convergence_detection.run_diffsel" [
+    docker env (
+      and_list [
+        mkdir_p dest;
+        cmd "cp" [dep phy_n; dest_ali]; (* required dep to link the file in the env *)
+        cmd "cp" [dep tree; dest_tree]; (* required dep to link the file in the env *)
+        cmd "cp" [dep phy_n; tmp_ali]; (* required dep to link the file in the env *)
+        cmd "cp" [dep tree; tmp_tree]; (* required dep to link the file in the env *)
+        cmd "/diffsel/_build/diffsel" [
+                   opt "-t" ident tmp_tree;
+                   opt "-d" ident tmp_ali ;
+                   opt "-ncond"  int 2 ;
+                   opt "-x" seq [ int w_every; string " "; int n_cycles];
+                   ident chainname ;
+        ];
+      ]
+    )
+  ]
+
+let diffsel_selector run_diffsel : text_file workflow =
+  let env = docker_image ~account:"vlanore" ~name:"diffsel" ~tag:"v1.0" () in
   let package = tmp // "diffsel_script_utils.py" in
   let script = tmp // "diffsel_analyze_result.py" in
+  let tmp_tree = tmp // "myrun.tree" in
+  let tmp_ali = tmp // "myrun.ali" in
+  let dep_tree = (dep run_diffsel) // "myrun.tree" in
+  let dep_ali = (dep run_diffsel) // "myrun.ali" in
   let chainname = (dep run_diffsel) // "myrun" in
   let out = dest in
   workflow ~descr:"convergence_detection.parse_diffsel" [
@@ -68,8 +90,9 @@ let diffsel ~(phy_n:nucleotide_phylip workflow) ~(tree: _ workflow) ~(w_every:in
         mkdir_p tmp ;
         cd tmp ;
         cmd "ls" [dep run_diffsel];
-        cmd "ls" [dep phy_n]; (* required dep to link the file in the env *)
-        cmd "ls" [dep tree]; (* required dep to link the file in the env *)
+
+        cmd "cp" [dep_ali; tmp_ali]; (* required dep to link the file in the env *)
+        cmd "cp" [dep_tree; tmp_tree]; (* required dep to link the file in the env *)
 
         (*python diffsel_analyze_result.py [-r /path/to/readdiffsel] [-o output_file] chainname *)
         cmd "cp"  [ file_dump (string Scripts.diffsel_script_utils) ; package] ;
@@ -91,9 +114,9 @@ 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
+      | Pcoc -> pcoc_selector res.det_result
+      | Pcoc_gamma -> pcoc_selector res.det_result
+      | Diffsel -> diffsel_selector res.det_result
     in
     let opt = match def_meth with
       | Pcoc -> string "--pcoc"

lib/convergence_detection.mli

@@ -33,6 +33,10 @@ val pcoc :
   tree : _ workflow ->
   (*[`pcoc]*) det_out directory workflow
 
+val pcoc_selector :
+    (*[`pcoc]*) det_out directory workflow ->
+    text_file workflow
+
 val diffsel :
   phy_n : nucleotide_phylip workflow ->
   tree : _ workflow ->
@@ -40,6 +44,9 @@ val diffsel :
   n_cycles: int ->
   (*[`diffsel]*) det_out directory workflow
 
+val diffsel_selector :
+    (*[`diffsel]*) det_out directory workflow ->
+    text_file workflow
 
 val merge_results :
   res_by_tools : det_result list ->

lib/pipeline.ml

@@ -105,11 +105,17 @@ let repo_of_dataset_results_l ~dataset_results_l =
           in
         let repo_d = Repo.shift "Detection_tools" (Repo.shift det_meth_prefix (
           Repo.[
-            match det_meth with
-              | Pcoc -> item ["pcoc.results.tsv"] w
-              | Pcoc_gamma -> item ["pcoc_gamma.results.tsv"] w
-              | Diffsel -> item ["diffsel.results.tsv"] w
-          ]
+            [match det_meth with
+              | Pcoc -> item ["pcoc.results.tsv"] (pcoc_selector w)
+              | Pcoc_gamma -> item ["pcoc_gamma.results.tsv"] (pcoc_selector w)
+              | Diffsel -> item ["diffsel.results.tsv"] (diffsel_selector w)
+            ];
+            [match det_meth with
+              | Pcoc -> item ["raw_results"]  w
+              | Pcoc_gamma -> item ["raw_results"] w
+              | Diffsel -> item ["raw_results"] w
+           ] ;
+          ] |> List.concat
         ))
         in
         repo_d