refactoring: started toolbox library

toolbox will be used to separate simple ocaml functions from bistro
workflows.

lib/convergence_detection.ml

@@ -190,7 +190,8 @@ let recall_precision_curve table =
   ]
 
 let%workflow recall_precision_auc_table table =
-  let labels, scores_per_meth = Utils.result_table_parser [%path table] in
+  let module RT = Reviewphiltrans_toolbox.Result_table in
+  let { RT.labels ; scores_per_meth } = RT.of_file [%path table] in
   Array.map scores_per_meth ~f:(fun (meth, scores) ->
       let _, auc = Biocaml_unix.Bin_pred.recall_precision_curve ~labels ~scores in
       meth, auc

lib/dune

 (library
  (name reviewphiltrans)
- (libraries biocaml.ez bistro.bioinfo bistro.utils convdet gzt ocaml-r.graphics ocaml-r.grDevices)
+ (libraries bistro.bioinfo bistro.utils convdet gzt ocaml-r.graphics ocaml-r.grDevices reviewphiltrans_toolbox )
  (preprocess
   (pps ppx_jane ppx_csv_conv bistro.ppx ppx_here)))
 

lib/toolbox/dune

+(library
+ (name reviewphiltrans_toolbox)
+ (libraries biocaml.ez convdet gzt ocaml-r.graphics ocaml-r.grDevices)
+ (preprocess (pps ppx_jane)))

lib/toolbox/result_table.ml

+open Core_kernel
+
+type t = {
+  labels : bool array ;
+  scores_per_meth : (string * float array) array ;
+}
+
+let of_file fn =
+  let module D = Owl.Dataframe in
+  let df = D.of_csv fn in
+  let labels =
+    D.get_col_by_name df "Oracle"
+    |> D.unpack_int_series
+    |> Array.map ~f:(function
+        | 0 -> false
+        | 1 -> true
+        | _ -> assert false
+      )
+  in
+  let scores meth =
+    match D.get_col_by_name df meth with
+    | Float_Series xs -> Some xs
+    | Int_Series xs -> Some (Array.map xs ~f:Float.of_int)
+    | String_Series _ -> None (* has NA *)
+    | _ -> assert false
+  in
+  let methods =
+    D.get_heads df
+    |> Array.filter ~f:(fun m -> m <> "Oracle" && m <> "Sites")
+  in
+  let scores_per_meth =
+    Array.filter_map methods ~f:(fun m ->
+        Option.map (scores m) ~f:(fun c -> m, c)
+      )
+  in
+  { labels ; scores_per_meth }

lib/utils.ml

@@ -59,31 +59,3 @@ let%pworkflow recall_precision_curve ~labels ~oracle ~results : Bistro.pdf Bistr
   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_grDevices.dev_off ()
-
-let result_table_parser fn =
-  let module D = Owl.Dataframe in
-  let df = D.of_csv fn in
-  let labels =
-    D.get_col_by_name df "Oracle"
-    |> D.unpack_int_series
-    |> Array.map ~f:(function
-        | 0 -> false
-        | 1 -> true
-        | _ -> assert false
-      )
-  in
-  let scores meth =
-    match D.get_col_by_name df meth with
-    | Float_Series xs -> Some xs
-    | Int_Series xs -> Some (Array.map xs ~f:Float.of_int)
-    | String_Series _ -> None (* has NA *)
-    | _ -> assert false
-  in
-  let methods =
-    D.get_heads df
-    |> Array.filter ~f:(fun m -> m <> "Oracle" && m <> "Sites")
-  in
-  labels,
-  Array.filter_map methods ~f:(fun m ->
-      Option.map (scores m) ~f:(fun c -> m, c)
-    )