Utils: fixed recall_precision_curve

lib/utils.ml

@@ -34,23 +34,35 @@ let%pworkflow fasta_cappend fa1 fa2 =
   |> Out_channel.write_lines [%dest]
 
 let%pworkflow recall_precision_curve ~labels ~oracle ~results : pdf file =
+  let open Codepitk in
   let n = Array.length labels in
   let colors = Array.sub [|"blue";"green";"red";"tan1";"royalblue4";"darkorchid"|] ~pos:0 ~len:n in
   let labels = [%param labels] in
   let load_results fn =
-    let open Owl.Dataframe in
-    match get_col (of_csv ~sep:'\t' fn) 1 with
-    | Float_Series xs -> xs
-    | Int_Series xs -> Array.map xs ~f:Float.of_int
-    | _ -> assert false
+    let df = Dataframe.from_file fn  |> Rresult.R.failwith_error_msg in
+    match Dataframe.get_col df 1 with
+    | Some (Floats xs) -> Array.map xs ~f:Option.some
+    | Some (Float_opts xs) -> xs
+    | Some (Ints xs) -> Array.map xs ~f:(fun i -> Some (Float.of_int i))
+    | Some (Int_opts xs) -> Array.map xs ~f:(Option.map ~f:Float.of_int)
+    | _ -> failwith "expected a numeric column at pos 1"
   in
   let oracle =
     load_results [%path oracle]
-    |> Array.map ~f:Float.(( <> ) 0.)
+    |> Array.map ~f:(Option.map ~f:Float.(( <> ) 0.))
   in
   let results = List.map ~f:load_results [%eval Bistro.Workflow.path_list results] in
+  let filter_table x y =
+    Array.filter_mapi x ~f:(fun i x_i ->
+        match x_i, y.(i) with
+        | Some _, Some _ -> x_i
+        | None, _ | _, None -> None
+      )
+  in
   let result_curves =
     List.map results ~f:(fun scores ->
+        let scores = filter_table scores oracle
+        and oracle = filter_table oracle scores in
         Biocaml_unix.Bin_pred.recall_precision_curve ~scores ~labels:oracle
         |> fst
         |> Array.unzip
@@ -59,7 +71,7 @@ let%pworkflow recall_precision_curve ~labels ~oracle ~results : pdf file =
   OCamlR_grDevices.pdf [%dest] ;
   OCamlR_graphics.plot ~xlab:"Recall" ~ylab:"Precision" ~xlim:(0., 1.) ~ylim:(0., 1.) ~x:[||] ~y:[||] () ;
   List.iteri result_curves ~f:(fun i (x, y) -> OCamlR_graphics.lines ~col:colors.(i) ~x ~y ()) ;
-  OCamlR_graphics.legend ~pch:[|19|] ~col:colors `bottomleft labels ;
+  OCamlR_graphics.legend ~pch:[|19|] ~col:colors `topright labels ;
   OCamlR_grDevices.dev_off ()
 
 let%pworkflow [@version 2] newick_of_nhx (t : nhx file) : newick file =