added diffseldsparse to new benchmark

lib/convergence_detection.ml

@@ -78,13 +78,14 @@ let merge_results ?fna_infos ~(res_by_tools : result list) () : text_file pworkf
     ] ;
   ]
 
-let merge_result_tables ?fna_infos ?oracle ?multinomial ?diffsel () : text_file pworkflow =
+let merge_result_tables ?fna_infos ?oracle ?multinomial ?diffsel ?diffseldsparse () : text_file pworkflow =
   Workflow.shell ~descr:"convergence_detection.merge_results" [
     cmd "python" ~img:Env.env_py [
       file_dump (string Scripts.merge_det_results) ;
       opt "-o" ident dest ;
       option (opt "--multinomial" dep) multinomial ;
       option (opt "--diffsel" dep) diffsel ;
+      option (opt "--diffseldsparse" dep) diffseldsparse ;
       option (opt "--oracle" dep) oracle ;
       option (opt "--fna_infos" dep) fna_infos;
     ] ;

lib/convergence_detection.mli

@@ -37,6 +37,7 @@ val merge_result_tables :
   ?oracle:text_file pworkflow ->
   ?multinomial:text_file pworkflow ->
   ?diffsel:text_file pworkflow ->
+  ?diffseldsparse:text_file pworkflow ->
   unit ->
   text_file pworkflow
 

lib/diffseldsparse.ml

@@ -148,3 +148,17 @@ let posterior_probabilities run_diffseldsparse : text_file pworkflow =
       ]
     )
   ]
+
+let%pworkflow results dir =
+  Filename.concat [%path dir] "myrun_1.sitepp"
+  |> In_channel.read_lines
+  |> Fn.flip List.drop 1
+  |> List.map ~f:(String.split ~on:'\t')
+  |> List.map ~f:(function
+      | site :: pp :: _ ->
+        Int.of_string site + 1, Float.of_string pp /. 100.
+      | _ -> failwith "parsing readdiffseldsparse results"
+    )
+  |> List.map ~f:(fun (site, pp) -> sprintf "%d\t%f" site pp)
+  |> (fun xs -> "Sites\tDiffseldsparse" :: xs)
+  |> Out_channel.write_lines [%dest]

lib/diffseldsparse.mli

@@ -18,8 +18,12 @@ val posterior_probabilities :
 
 val readdiffseldsparse :
   [`diffseldsparse] dworkflow ->
-  text_file pworkflow
+  [`readdiffseldsparse] dworkflow
 
 val check_conv :
   [`diffseldsparse] dworkflow ->
   directory pworkflow
+
+val results :
+  [`readdiffseldsparse] dworkflow ->
+  text_file pworkflow

lib/pipeline2.ml

@@ -109,6 +109,7 @@ let diffseldsparse ?pi ?shiftprob ?eps d =
     ~n_cycles:50
     ()
   |> Diffseldsparse.readdiffseldsparse
+  |> Diffseldsparse.results
 
 let alignment_plot d =
   Convergence_detection.plot_convergent_sites
@@ -133,6 +134,7 @@ let benchmark d =
   Convergence_detection.merge_result_tables
     ~multinomial:(multinomial d)
     ~diffsel:(diffsel d)
+    ~diffseldsparse:(diffseldsparse ~eps:1e-3 d)
     ~oracle:(oracle d)
     ()
   |> Convergence_detection.recall_precision_curve

lib/scripts/merge_det_results.py

@@ -60,6 +60,8 @@ availableOptions.add_argument('--diffsel', type=str,
                    help="Diffsel output name", default = None)
 availableOptions.add_argument('--diffsel_bis', type=str,
                    help="Diffsel_bis output name", default = None)
+availableOptions.add_argument('--diffseldsparse', type=str,
+                   help="Diffseldsparse output name", default = None)
 availableOptions.add_argument('--identical_LG', type=str,
                    help="Identical_LG output name", default = None)
 availableOptions.add_argument('--identical_WAG', type=str,
@@ -89,6 +91,7 @@ df_pcoc = pd.DataFrame()
 df_pcoc_gamma = pd.DataFrame()
 df_pcoc_C60 = pd.DataFrame()
 df_diffsel = pd.DataFrame()
+df_diffseldsparse = pd.DataFrame()
 df_diffsel_bis = pd.DataFrame()
 df_identical_LG = pd.DataFrame()
 df_identical_WAG = pd.DataFrame()
@@ -127,6 +130,9 @@ if args.diffsel_bis :
 if args.diffsel :
   df_diffsel = pd.read_csv(args.diffsel, sep="\t")
 
+if args.diffseldsparse :
+  df_diffseldsparse = pd.read_csv(args.diffseldsparse, sep="\t")
+  
 if args.oracle:
   df_oracle = pd.read_csv(args.oracle, sep="\t")
   df_oracle.rename(columns={'Oracle' : 'Oracle'}, inplace=True)
@@ -172,7 +178,7 @@ if args.fna_infos :
   #df_fna_infos = df_fna_infos[['Sites','P_distance']]
 
 df_list = [df for df in [df_pcoc, df_pcoc_gamma, df_pcoc_C60,
-                         df_diffsel, df_diffsel_bis,
+                         df_diffsel, df_diffsel_bis, df_diffseldsparse,
                          df_identical_LG, df_identical_WAG,
                          df_topological_LG, df_topological_WAG,
                          df_tdg09,