orthomam: direct access to prestin

lib/orthomam.ml

@@ -11,6 +11,11 @@ type alignment = Alignment of Orthomam_db.t * Orthomam_db.alignment
 let alignments_of_db db =
   List.map (Orthomam_db.list_alignments db) ~f:(fun ali -> Alignment (db, ali))
 
+let family_id_of_alignment (alignment : Orthomam_db.alignment) =
+  Filename.basename (alignment :> string)
+  |> Fn.flip Filename.chop_suffix "_NT.fasta.ali"
+  |> String.chop_prefix_exn ~prefix:"all"
+
 let%pworkflow fasta_of_phylip ali =
   let ali =
     [%path ali]
@@ -30,10 +35,7 @@ type query = {
 let query (Alignment (db, alignment)) ~convergent_species =
   { db ; alignment ; convergent_species }
 
-let family_id q =
-  Filename.basename (q.alignment :> string)
-  |> Fn.flip Filename.chop_suffix "_NT.fasta.ali"
-  |> String.chop_prefix_exn ~prefix:"all"
+let family_id q = family_id_of_alignment q.alignment
 
 let%pworkflow remove_unobserved_sequences_from_alignment phylip : phylip file =
   let phylip = [%path phylip] in
@@ -395,6 +397,15 @@ let species_with_echolocation = [
   (* "Rhinolophus ferrumequinum" ; in ensembl but not in orthomam *)
 ]
 
+let prestin db =
+  let alignments = alignments_of_db db in
+  let prestin_alignment = List.find_exn alignments ~f:(fun (Alignment (_, al)) ->
+      String.equal (family_id_of_alignment al) "ENSG00000170615_SLC26A5"
+    )
+  in
+  let convergent_species = Workflow.data species_with_echolocation in
+  query ~convergent_species prestin_alignment
+
 let%pworkflow convergence_species_tree_pdf ~convergent_species db =
   let convergent_species = [%param convergent_species] in
   let tree_path = [%path estimated_amino_acid_tree db] in

lib/orthomam.mli

@@ -28,6 +28,8 @@ include Detection_pipeline.S with type query := query
 
 val species_with_echolocation : string list
 
+val prestin : Orthomam_db.t -> query
+
 val ensembl_tree : nhx file
 
 val compare_tree_branch_lengths :