orthomam app

app/dune

@@ -13,3 +13,11 @@
  (libraries reviewphiltrans)
  (preprocess
   (pps ppx_jane)))
+
+(executable
+ (name orthomam_app)
+ (public_name orthomam_convergence)
+ (modules orthomam_app)
+ (libraries reviewphiltrans)
+ (preprocess
+  (pps ppx_jane)))

app/orthomam_app.ml

+open Reviewphiltrans
+
+module Top = Bistro_utils.Toplevel_eval.Make(struct let np = 8 let mem = 10 end)()
+
+let () =
+  Reviewphiltrans_toolbox.Orthomam_db.make "/disk/data/omm"
+  |> Orthomam.(
+      site_ranking
+        ~subset:5
+        ~convergent_species:species_with_echolocation
+        ~meth:`tdg09
+    )
+  |> Top.eval
+  |> List.length
+  |> print_int

lib/orthomam.ml

@@ -32,7 +32,7 @@ let query (Alignment (db, alignment)) ~convergent_species =
 
 let family_id q =
   Filename.basename q.alignment
-  |> Filename.chop_suffix "_NT.fasta.ali"
+  |> Fn.flip Filename.chop_suffix "_NT.fasta.ali"
   |> String.chop_prefix_exn ~prefix:"all"
 
 let%pworkflow remove_unobserved_sequences_from_alignment phylip : phylip file =
@@ -581,13 +581,16 @@ let%workflow ranking_of_results ~alignment_ids ~convergent_species (alignments :
   in
   List.map top_sites ~f:annotate_site
 
-let site_ranking ~meth ~convergent_species db =
+let site_ranking ?subset ~meth ~convergent_species db =
   let detection_method, column_label = match meth with
     | `multinomial_asymptotic_lrt ->
       multinomial_asymptotic_lrt, "1MinusLRT"
+    | `tdg09 ->
+      tdg09, "Tdg09_1MinusFDR"
   in
   let queries =
     alignments_of_db db
+    |> (match subset with None -> Fn.id | Some n -> Fn.flip List.take n)
     |> List.map ~f:(query ~convergent_species:(Workflow.data convergent_species))
   in
   let alignment_ids = List.map queries ~f:family_id in

lib/orthomam.mli

@@ -88,7 +88,8 @@ module Candidate_site : sig
 end
 
 val site_ranking :
-  meth:[`multinomial_asymptotic_lrt] ->
+  ?subset:int ->
+  meth:[`multinomial_asymptotic_lrt | `tdg09] ->
   convergent_species:string list ->
   Orthomam_db.t ->
   Candidate_site.t list workflow