Run: report for multinomial

lib/dataset.ml

@@ -56,7 +56,7 @@ open File_formats
 module New_API = struct
   type t = {
     tree : nhx file ;
-    nucleotide_alignments : nucleotide_fasta file list ;
+    nucleotide_alignments : (string * nucleotide_fasta file) list ;
     convergent_species : string list workflow ;
   }
 
@@ -107,14 +107,20 @@ module New_API = struct
 
   module Query = struct
     type dataset = t
-    type t = dataset * nucleotide_fasta file
+    type t = {
+      dataset : dataset ;
+      alignment_descr : string ;
+      alignment : nucleotide_fasta file ;
+    }
 
-    let nucleotide_alignment = snd
+    let nucleotide_alignment q = q.alignment
 
-    let tree ~branch_length_unit:_ (d, fa) =
-      clip_tree_on_alignment d.tree fa
+    let tree ~branch_length_unit:_ q =
+      clip_tree_on_alignment q.dataset.tree q.alignment
   end
 
-  let queries d =
-    List.map d.nucleotide_alignments ~f:(fun al -> d, al)
+  let queries dataset =
+    List.map dataset.nucleotide_alignments ~f:(fun (alignment_descr, alignment) ->
+        { Query.dataset ; alignment_descr ; alignment }
+      )
 end

lib/run.ml

@@ -2,6 +2,31 @@ open Core
 open Bistro
 open Bistro_utils
 
+let candidate_site_report sites =
+  let f = fun%workflow dest ->
+    let module N = Codepitk.Note in
+    let module DF = Codepitk.Dataframe in
+    let module CS = Codepitk.Candidate_site in
+    let sites = Array.of_list [%eval sites] in
+    let df =
+      DF.make [
+        "Alignment ID", DF.String_opts (Array.map sites ~f:(fun s -> s.CS.alignment_id)) ;
+        "Position", DF.Int_opts (Array.map sites ~f:(fun s -> s.CS.pos)) ;
+        "Score", DF.Float_opts (Array.map sites ~f:(fun s -> s.CS.score)) ;
+      ]
+      |> Rresult.R.failwith_error_msg
+    in
+    let index = N.make ~title:"Candidate sites" N.[
+        dataframe df ;
+      ]
+    in
+    let path fn = Filename.concat dest fn in
+    Unix.mkdir_p dest ;
+    N.to_html index (path "index.html")
+    |> Rresult.R.failwith_error_msg
+  in
+  Workflow.path_plugin ~descr:"codepi.run.candidate_site_report" f
+
 module Dataset = Dataset.New_API
 
 type t = {
@@ -21,19 +46,29 @@ let dataset run =
     Sys.readdir run.alignment_dir
     |> Array.to_list
     |> List.map ~f:(fun ali ->
+        ali,
         Workflow.input (Filename.concat run.alignment_dir ali)
       )
   in
   let convergent_species = convergent_species_workflow run in
-  { Dataset.tree ; nucleotide_alignments ; convergent_species }
+  Dataset.make ~tree ~nucleotide_alignments ~convergent_species
 
 module Pipeline = Detection_pipeline.Make(Dataset.Query)
 
 let repo run =
   let d = dataset run in
-  let q = d, List.hd_exn d.nucleotide_alignments in
+  let q = List.hd_exn (Dataset.queries d) in
+  let multinomial_ranking =
+    Pipeline.ranking
+      ~query_descr:(fun q -> Some q.Dataset.Query.alignment_descr)
+      ~meth:Pipeline.multinomial_asymptotic_lrt
+      ~column_label:"Multinomial_1mp"
+      ~convergent_species:d.convergent_species
+      (Dataset.queries d)
+  in
   Repo.[
     item ["multinomial.cpt"] (Pipeline.multinomial_asymptotic_lrt q) ;
+    item ["multinomial_report"] (candidate_site_report multinomial_ranking) ;
   ]
 
 let main