More precise type for detection method outputs

app/lmm_benchmark.ml

@@ -51,7 +51,7 @@ let orthomam_echolocation = {
 }
 
 type detection_method = {
-  result : Pipeline.query -> text file ;
+  result : Pipeline.query -> cpt file ;
   col : int ;
   label : string ;
   requires_rooted_tree : bool ;

lib/convergence_detection.ml

 open Core
 open Bistro.Shell_dsl
 open Bistro
+open File_formats
 
 type result = [
   | `Pcoc of [`pcoc] directory
@@ -12,7 +13,7 @@ type result = [
   | `Topological_LG of [`topological] directory
   | `Topological_WAG of [`topological] directory
   | `Tdg09 of [`tdg09] directory
-  | `Multinomial of text file
+  | `Multinomial of cpt file
   | `Msd of [`msd] directory * float
 ]
 
@@ -34,11 +35,11 @@ type dataset_res = {
   tree_prefix : string ;
   dataset : Dataset.t ;
   res_by_tools: result list ;
-  merged_results : text file ;
+  merged_results : cpt file ;
   plot_merged_results : svg file ;
 }
 
-let merge_results ?fna_infos ~(res_by_tools : result list) () : text file =
+let merge_results ?fna_infos ~(res_by_tools : result list) () : cpt file =
   let command = List.map res_by_tools ~f:(fun res ->
       let w = match res with
         | `Pcoc d -> Pcoc.results d
@@ -78,7 +79,7 @@ let merge_results ?fna_infos ~(res_by_tools : result list) () : text file =
     ] ;
   ]
 
-let merge_result_tables ?fna_infos ?oracle ?multinomial ?tdg09 ?identical ?topological ?pcoc ?pcoc_v2 ?pcoc_pcp ?diffsel ?diffseldsparse () : text file =
+let merge_result_tables ?fna_infos ?oracle ?multinomial ?tdg09 ?identical ?topological ?pcoc ?pcoc_v2 ?pcoc_pcp ?diffsel ?diffseldsparse () : cpt file =
   Workflow.shell ~descr:"convergence_detection.merge_results" ~img:Env.env_py [
     cmd "python" [
       file_dump (string Scripts.merge_det_results) ;

lib/convergence_detection.mli

@@ -11,7 +11,7 @@ type result = [
   | `Topological_LG of [`topological] directory
   | `Topological_WAG of [`topological] directory
   | `Tdg09 of [`tdg09] directory
-  | `Multinomial of text file
+  | `Multinomial of cpt file
   | `Msd of [`msd] directory * float
 ]
 
@@ -22,7 +22,7 @@ type dataset_res = {
   tree_prefix : string ;
   dataset : Dataset.t ;
   res_by_tools: result list ;
-  merged_results : text file ;
+  merged_results : cpt file ;
   plot_merged_results : svg file
 }
 
@@ -30,22 +30,22 @@ val merge_results :
   ?fna_infos:text file ->
   res_by_tools : result list ->
   unit ->
-  text file
+  cpt file
 
 val merge_result_tables :
   ?fna_infos:text file ->
-  ?oracle:text file ->
-  ?multinomial:text file ->
-  ?tdg09:text file ->
-  ?identical:text file ->
-  ?topological:text file ->
-  ?pcoc:text file ->
-  ?pcoc_v2:text file ->
-  ?pcoc_pcp:text file ->
-  ?diffsel:text file ->
-  ?diffseldsparse:text file ->
+  ?oracle:cpt file ->
+  ?multinomial:cpt file ->
+  ?tdg09:cpt file ->
+  ?identical:cpt file ->
+  ?topological:cpt file ->
+  ?pcoc:cpt file ->
+  ?pcoc_v2:cpt file ->
+  ?pcoc_pcp:cpt file ->
+  ?diffsel:cpt file ->
+  ?diffseldsparse:cpt file ->
   unit ->
-  text file
+  cpt file
 
 val plot_merge_results :
   ? t_choices : text file ->
@@ -53,27 +53,27 @@ val plot_merge_results :
   res_by_tools : result list ->
   tree:nhx file ->
   faa:aminoacid_fasta file ->
-  tsv:text file ->
+  tsv:cpt file ->
   unit ->
   svg file
 
 val plot_convergent_sites :
   ?plot_all_sites:bool ->
   alignment:aminoacid_fasta file ->
-  detection_results:text file ->
+  detection_results:cpt file ->
   tree:nhx file ->
   unit ->
   svg file
 
 val recall_precision_curve :
-  text file ->
+  cpt file ->
   svg file
 
 val oracle :
   n_h0:int ->
   n_ha:int ->
-  text file
+  cpt file
 
 val recall_precision_auc_table :
-  text file ->
+  cpt file ->
   (string * float) list workflow

lib/detection_pipeline.ml

@@ -27,45 +27,45 @@ module type S = sig
 
   val dnds_tree : query -> text file
 
-  val identical : query -> text file
+  val identical : query -> cpt file
 
-  val topological : query -> text file
+  val topological : query -> cpt file
 
-  val multinomial : query -> text file
+  val multinomial : query -> cpt file
 
-  val multinomial_simulation_lrt : query -> text file
+  val multinomial_simulation_lrt : query -> cpt file
 
-  val multinomial_simulation_sparse : query -> text file
+  val multinomial_simulation_sparse : query -> cpt file
 
-  val multinomial_asymptotic_lrt : query -> text file
+  val multinomial_asymptotic_lrt : query -> cpt file
 
-  val multinomial_asymptotic_sparse : query -> text file
+  val multinomial_asymptotic_sparse : query -> cpt file
 
-  val tdg09 : query -> text file
+  val tdg09 : query -> cpt file
 
-  val failsafe_tdg09 : query -> text file
+  val failsafe_tdg09 : query -> cpt file
 
-  val pcoc : ?gamma:bool -> ?ncat:int -> query -> text file
+  val pcoc : ?gamma:bool -> ?ncat:int -> query -> cpt file
 
   val pcoc_v2 :
-    ?gamma:bool -> ?aa_profiles:Pcoc.aa_profiles -> query -> text file
+    ?gamma:bool -> ?aa_profiles:Pcoc.aa_profiles -> query -> cpt file
 
   val gemma :
     query ->
     lmm_test:[ `All | `LRT | `Score | `Wald ] ->
     relatedness_mode:[ `Centered | `Standardized ] ->
-    text file
+    cpt file
 
-  val inhouse_lmm : query -> text file
+  val inhouse_lmm : query -> cpt file
 
-  val diffsel : query -> text file
+  val diffsel : query -> cpt file
 
   val diffseldsparse :
     ?pi:float ->
     ?shiftprob:float * float ->
     ?eps:float ->
     query ->
-    text file
+    cpt file
 
   val view_site :
     query -> convergent_species:string list -> site_pos:int -> pdf file

lib/detection_pipeline.mli

@@ -27,45 +27,45 @@ module type S = sig
 
   val dnds_tree : query -> text file
 
-  val identical : query -> text file
+  val identical : query -> cpt file
 
-  val topological : query -> text file
+  val topological : query -> cpt file
 
-  val multinomial : query -> text file
+  val multinomial : query -> cpt file
 
-  val multinomial_simulation_lrt : query -> text file
+  val multinomial_simulation_lrt : query -> cpt file
 
-  val multinomial_simulation_sparse : query -> text file
+  val multinomial_simulation_sparse : query -> cpt file
 
-  val multinomial_asymptotic_lrt : query -> text file
+  val multinomial_asymptotic_lrt : query -> cpt file
 
-  val multinomial_asymptotic_sparse : query -> text file
+  val multinomial_asymptotic_sparse : query -> cpt file
 
-  val tdg09 : query -> text file
+  val tdg09 : query -> cpt file
 
-  val failsafe_tdg09 : query -> text file
+  val failsafe_tdg09 : query -> cpt file
 
-  val pcoc : ?gamma:bool -> ?ncat:int -> query -> text file
+  val pcoc : ?gamma:bool -> ?ncat:int -> query -> cpt file
 
   val pcoc_v2 :
-    ?gamma:bool -> ?aa_profiles:Pcoc.aa_profiles -> query -> text file
+    ?gamma:bool -> ?aa_profiles:Pcoc.aa_profiles -> query -> cpt file
 
   val gemma :
     query ->
     lmm_test:[ `All | `LRT | `Score | `Wald ] ->
     relatedness_mode:[ `Centered | `Standardized ] ->
-    text file
+    cpt file
 
-  val inhouse_lmm : query -> text file
+  val inhouse_lmm : query -> cpt file
 
-  val diffsel : query -> text file
+  val diffsel : query -> cpt file
 
   val diffseldsparse :
     ?pi:float ->
     ?shiftprob:float * float ->
     ?eps:float ->
     query ->
-    text file
+    cpt file
 
   val view_site :
     query -> convergent_species:string list -> site_pos:int -> pdf file

lib/diffsel.ml

@@ -106,7 +106,7 @@ let check_conv run_diffsel : [`diffsel_check_conv] directory =
     ]
   ]
 
-let selector run_diffsel : text file =
+let selector run_diffsel : cpt file =
   let env = Env.env_diffsel in
   let package = tmp // "diffsel_script_utils.py" in
   let script = tmp // "diffsel_analyze_result.py" in

lib/diffsel.mli

@@ -13,7 +13,7 @@ val diffsel :
 
 val selector :
   [`diffsel] directory ->
-  text file
+  cpt file
 
 val check_conv :
   [`diffsel] directory ->

lib/diffseldsparse.ml

@@ -117,7 +117,7 @@ let readdiffseldsparse run =
     ]
   ]
 
-let posterior_probabilities run_diffseldsparse : text file =
+let posterior_probabilities run_diffseldsparse : cpt file =
   let tmp_tree = tmp // "myrun.tree" in
   let tmp_ali = tmp // "myrun.ali" in
   let dep_tree = (dep run_diffseldsparse) // "myrun.tree" in

lib/diffseldsparse.mli

@@ -14,7 +14,7 @@ val diffseldsparse :
 
 val posterior_probabilities :
   [`diffseldsparse] directory ->
-  text file
+  cpt file
 
 val readdiffseldsparse :
   [`diffseldsparse] directory ->
@@ -26,4 +26,4 @@ val check_conv :
 
 val results :
   [`readdiffseldsparse] directory ->
-  text file
+  cpt file

lib/file_formats.ml

@@ -54,3 +54,10 @@ class type rds = object
   inherit binary_file
   method format : [`rds]
 end
+
+(** Convergence Prediction Table *)
+class type cpt = object
+  inherit tsv
+  method header : [`Yes]
+  method fields : [`Site_then_scores]
+end

lib/gemma.mli

@@ -33,4 +33,4 @@ val univariate_lmm :
   univariate_lmm_output file
 
 val result_table_of_output :
-  aminoacid_fasta file -> univariate_lmm_output file -> text file
+  aminoacid_fasta file -> univariate_lmm_output file -> cpt file

lib/identical.ml

@@ -105,5 +105,5 @@ let identical ?(descr="") ~(tree_id:_ file) ~(tree_sc:_ file) ~(faa:aminoacid_fa
     ]
   ]
 
-let results run_identical : text file =
+let results run_identical : cpt file =
   Workflow.select run_identical ["out1.tsv"]

lib/inhouse_lmm.mli

@@ -4,4 +4,4 @@ open File_formats
 val test :
   aminoacid_fasta file ->
   nhx file ->
-  text file
+  cpt file

lib/msd.ml

@@ -42,7 +42,7 @@ let msd ?(descr="") ~e ~(faa : aminoacid_fasta file) ~(tree_sc : _ file) : [`msd
     ];
   ]
 
-let results run_msd : text file =
+let results run_msd : cpt file =
   Workflow.shell ~descr:"convergence_detection.parse_msd" ~img [
     cmd "python" [
       file_dump (string Scripts.parse_output_msd) ;

lib/multinomial.ml

@@ -3,7 +3,7 @@ open Bistro
 open Bistro.Shell_dsl
 open File_formats
 
-let multinomial ?(descr="") ~(tree_sc:_ file) ~(faa:aminoacid_fasta file) () : text file =
+let multinomial ?(descr="") ~(tree_sc:_ file) ~(faa:aminoacid_fasta file) () : cpt file =
   let img = Env.env_py in
   Workflow.shell ~descr:("calc_multinomial."^descr) ~img [
     cmd "python" [
@@ -14,7 +14,7 @@ let multinomial ?(descr="") ~(tree_sc:_ file) ~(faa:aminoacid_fasta file) () : t
     ]
   ]
 
-let%pworkflow multinomial_ocaml_implementation ~meth ~(tree_sc:_ file) ~(faa:aminoacid_fasta file) (* : text file *) =
+let%pworkflow multinomial_ocaml_implementation ~meth ~(tree_sc:_ file) ~(faa:aminoacid_fasta file) (* : cpt file *) =
   let open Phylogenetics in
   let open Phylogenetics_convergence in
   let module MT = Multinomial_test in

lib/pcoc.ml

@@ -20,7 +20,7 @@ let pcoc ?(descr = "") ?plot_complete ?gamma ?catx_est ?max_gap_per_pos ?max_gap
     ]
   ]
 
-let results run_pcoc : text file =
+let results run_pcoc : cpt file =
   let out_pcoc = dep run_pcoc // "RUN*/*.results.tsv" in
   Workflow.shell ~descr:"convergence_detection.selector_pcoc" [
     cmd "cp" [out_pcoc; ident dest] ;

lib/pcoc.mli

@@ -15,7 +15,7 @@ val pcoc :
 
 val results :
   [`pcoc] directory ->
-  text file
+  cpt file
 
 type aa_profiles = 
   [`C10 |

lib/real_dataset.mli

@@ -27,7 +27,7 @@ val make :
 val result_table :
   meth list ->
   t ->
-  (string * text file) list
+  (string * cpt file) list
 
 val repo :
   meth list ->

lib/simulation_pipeline.mli

@@ -50,10 +50,10 @@ module Mutsel : sig
   val benchmark_statistics :
     Codepitk.Simulator.Site_independent_mutsel.simulation workflow ->
     labels:string list ->
-    results:(text file * int) list ->
+    results:(cpt file * int) list ->
     binary_file file
 
-  (* val benchmark : t -> (t -> text file) list -> benchmark workflow
+  (* val benchmark : t -> (t -> cpt file) list -> benchmark workflow
    *
    * val rds_of_benchmark : benchmark workflow -> rds file *)
 
@@ -103,7 +103,7 @@ module Bppseqgen : sig
 
   include Detection_pipeline.S with type query := t
 
-  val oracle : t -> text file
+  val oracle : t -> cpt file
 
   val alignment_plot : t -> svg file
 

lib/tamuri.ml

@@ -38,7 +38,7 @@ let tdg09  ?(descr="") ~(faa:aminoacid_fasta file) ~(tree:_ file) () : [`tdg09]
     ]
   ]
 
-let results run_tdg09 : text file =
+let results run_tdg09 : cpt file =
   let tdg09_out = Workflow.select run_tdg09 [ "tdg09.yaml" ] in
   Workflow.shell ~descr:"convergence_detection.parse_tdg09" ~img [
     cmd "python" [