orthomam: pdf repr of convergence in tree

lib/dune

 (library
  (name reviewphiltrans)
- (libraries core biotk biotope bistro.utils ocaml-r.graphics ocaml-r.grDevices phylogenetics.convergence reviewphiltrans_toolbox)
+ (libraries core biotk biotope bistro.utils gsl ocaml-r.graphics ocaml-r.grDevices phylogenetics.convergence reviewphiltrans_toolbox)
  (preprocess
   (pps ppx_jane ppx_csv_conv bistro.ppx ppx_here)))
 

lib/orthomam.ml

@@ -424,8 +424,65 @@ include Q
 include Detection_pipeline.Make(Q)
 
 let species_with_echolocation = [
-  "Rhinolophus ferrumequinum" ;
-  "Tursiops_truncatus" ;
+  "Orcinus_orca" ;
   "Myotis_lucifugus" ;
+  "Eptesicus_fuscus" ;
+  "Tursiops_truncatus" ;
+  "Miniopterus_natalensis" ;
+  (* "Pteropus_vampyrus" ; unsure, i'd say no *)
+  (* "Pteropus alecto" ; unsure, i'd say no *)
+  "Lipotes_vexillifer" ;
+  "Hipposideros_armiger" ;
   "Physeter_catodon" ;
+  "Rousettus_aegyptiacus" ; (* unsure, i'd say yes *)
+  "Myotis_brandtii" ;
+  "Balaenoptera_acutorostrata_scammoni" ;
+  "Delphinapterus_leucas" ;
+  "Myotis_davidii" ;
+  (* "Rhinolophus ferrumequinum" ; in ensembl but not in orthomam *)
 ]
+
+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
+  let open Phylogenetics in
+  let convergent_species = String.Set.of_list convergent_species in
+  let render_tree (tree_or_branch : Newick.t) =
+    let module P = Biotk.Phylo_tree_draw in
+    let rec of_tree : Newick.tree -> P.tree = function
+      | Leaf ni ->
+        let col, text =
+          match ni.name with
+          | Some name ->
+            let col =
+              if String.Set.mem convergent_species name
+              then Gg.Color.red else Gg.Color.black
+            in
+            col, String.tr ~target:'_' ~replacement:' ' name
+          | None -> Gg.Color.black, ""
+        in
+        P.leaf ~col ~style:`oblique text
+      | Node n ->
+        Phylogenetics.List1.map n.branches ~f:of_branch
+        |> Phylogenetics.List1.to_list
+        |> P.node
+    and of_branch : Newick.branch -> P.branch = fun (Branch b) ->
+      let len = Option.value_exn b.data.length in
+      let col =
+        match List.Assoc.find b.data.tags "Condition" ~equal:String.equal with
+        | Some "1" -> Gg.Color.red
+        | _ -> Gg.Color.black
+      in
+      P.branch ~col len (of_tree b.tip)
+    in
+    match tree_or_branch with
+    | Tree t -> P.draw_tree (of_tree t)
+    | Branch b -> P.draw_branch (of_branch b)
+  in
+  let tree_or_branch =
+    Newick.from_file tree_path
+    |> Newick.map_inner_tree ~f:(fun t -> tag_tree t convergent_species)
+  in
+  render_tree tree_or_branch
+  |> Biotk_croquis.Croquis.Layout.simple
+  |> Fn.flip Biotk_croquis.Croquis.Layout.render_pdf [%dest]

lib/orthomam.mli

@@ -66,3 +66,8 @@ val estimated_codon_tree :
 val estimated_amino_acid_tree :
   Orthomam_db.t ->
   newick file
+
+val convergence_species_tree_pdf :
+  convergent_species:string list ->
+  Orthomam_db.t ->
+  pdf file