fix errors comming from real data (indels)

lib/bppsuite.ml

@@ -118,6 +118,8 @@ let conf_file_bppseqman_fna2faa ~fna =
     assign "output.sequence.file" dest ;
     string {|alphabet=Codon(letter=DNA)
              genetic_code = Standard
+             input.sequence.remove_stop_codons = no
+             input.sequence.sites_to_use = all
              input.alignment = true
              sequence.manip = Translate
            |}
@@ -136,6 +138,8 @@ let conf_file_bppseqman_fa2phy ~fna =
     assign "output.sequence.file" dest ;
     assign "output.sequence.format" (string "Phylip") ;
     string {| input.alignment = true
+             input.sequence.remove_stop_codons = no
+             input.sequence.sites_to_use = all
              sequence.manip =
            |}
   ]

lib/identical.ml

@@ -12,15 +12,17 @@ let assign k v =
 
 let conf_file_bppml ~tree ~faa ~out ~config =
   seq ~sep:"\n" (
-    [ 
+    [
       assign "OUT" (out) ;
       assign "input.sequence.file" (dep faa) ;
       assign "alphabet" (string "Protein") ;
+      assign "input.sequence.remove_stop_codons" (string "no") ;
+      assign "input.sequence.sites_to_use" (string "all") ;
 
       assign "input.tree.file"  (dep tree) ;
       assign "init.tree" (string "user") ;
       assign "input.tree.format" (string "Nhx") ;
-      
+
       assign "optimization.topology" (string "false") ;
 
       assign "output.tree.file" (string "$(OUT)/tree.nhx") ;
@@ -48,11 +50,11 @@ let bppml ?(descr="") ~faa ~tree ~config : _ workflow =
       ]
     )
   ]
-  
+
 
 let conf_file_bppancestor ~tree ~faa ~out ~config =
   seq ~sep:"\n" (
-    [ 
+    [
       assign "OUT" (out) ;
       assign "input.sequence.file" (dep faa) ;
       assign "alphabet" (string "Protein") ;
@@ -60,7 +62,7 @@ let conf_file_bppancestor ~tree ~faa ~out ~config =
       assign "input.tree.file"  (dep tree) ;
       assign "init.tree" (string "user") ;
       assign "input.tree.format" (string "Nhx") ;
-      
+
       assign "optimization.topology" (string "false") ;
 
       assign "output.sequence.file" (string "$(OUT)/output_anc.fa") ;

lib/scripts/calc_topological.py

@@ -110,6 +110,7 @@ df_final["Sites"] = pd.to_numeric(df_final["Sites"].str.replace('[','').str.repl
 df_final["Topological"] = map(prob_ap,  df_final["lnL_conv"],df_final["lnL_noconv"])
 
 df_final = df_final[["Sites","Topological"]]
+
 #===================================================================================================
 # Create output files
 #===================================================================================================

lib/scripts/merge_det_results.py

@@ -132,9 +132,9 @@ df_list = [df for df in [df_pcoc, df_pcoc_gamma,
                         ] if not df.empty ]
 
 df_list_len = [df.shape[0] for df in df_list]
+
 if len(set(df_list_len)) != 1:
-    print("ERROR: all files have not the same number of rows")
-    sys.exit(1)
+    print("WARNING: all files have not the same number of rows")
 
 df_final = reduce(lambda x, y: pd.merge(x, y, on = 'Sites', how='outer'), df_list)
 

lib/topological.ml

@@ -16,6 +16,8 @@ let conf_file_bppml ~tree ~faa ~out ~config =
       assign "OUT" (out) ;
       assign "input.sequence.file" (dep faa) ;
       assign "alphabet" (string "Protein") ;
+      assign "input.sequence.remove_stop_codons" (string "no") ;
+      assign "input.sequence.sites_to_use" (string "all") ;
 
       assign "input.tree.file"  (dep tree) ;
       assign "init.tree" (string "user") ;