refactoring

lib/file_formats.ml

@@ -52,5 +52,5 @@ type ('a,'b) w_dataset =
 type dataset = {
   model_prefix: string ;
   tree_prefix : string ;
-  dataset : (raw_dataset,ready_dataset) w_dataset
+  ready_dataset : ready_dataset
   }

lib/pipeline.ml

@@ -11,6 +11,11 @@ let ready_dataset_of_raw_dataset raw_dataset =
     let faa = Bppsuite.fna2faa ~fna in
     { input_tree; tree_dataset ; fna; faa}
 
+let raw_dataset_of_ready_dataset ready_dataset =
+    let input_tree = ready_dataset.input_tree in
+    let fna = ready_dataset.fna in
+    { input_tree; fna}
+
 let repo_of_ready_dataset (rd : ready_dataset) =
   Repo.[
     item ["input_tree.nhx"] rd.input_tree ;
@@ -32,11 +37,14 @@ let repo_of_raw_dataset (raw_dataset:raw_dataset) =
   |> Repo.shift "raw_dataset"
 
 
-let repo_of_dataset_l dataset_l =
+let repo_of_dataset_l ~preview dataset_l =
   List.map dataset_l ~f:(fun dataset ->
     let model_prefix = dataset.model_prefix in
     let tree_prefix = dataset.tree_prefix in
-    let w_dataset = dataset.dataset in
+    let w_dataset =
+      if preview then
+        Ready_dataset dataset.ready_dataset
+      else Raw_dataset (raw_dataset_of_ready_dataset dataset.ready_dataset) in
     let repo_d = match w_dataset with
       | Ready_dataset d -> repo_of_ready_dataset d
       | Raw_dataset d ->  repo_of_raw_dataset d
@@ -53,14 +61,9 @@ let derive_from_model ~model ~tree ~tree_dataset ~tree_prefix ~preview =
   let tree = Tree_dataset.tree tree_dataset `Simulation in
   let fna = Bppsuite.bppseqgen ~nb_sites ~tree ~config in
   let faa = Bppsuite.fna2faa ~fna in
-  let raw_dataset = { input_tree = tree ; fna} in
   let ready_dataset = { input_tree = tree ; tree_dataset ; fna; faa} in
   let model_prefix = Convergence_hypothesis.string_of_model model in
-  if preview then
-    { model_prefix; tree_prefix; dataset = Ready_dataset ready_dataset }
-  else
-    { model_prefix; tree_prefix; dataset = Raw_dataset raw_dataset }
-
+    { model_prefix; tree_prefix; ready_dataset }
 
 let derive_from_tree ~tree_dir ~tree ~preview =
   let tree_prefix = Filename.chop_extension tree in
@@ -84,7 +87,7 @@ let main ~outdir ?(np = 2) ?(mem = 2) ~tree_dir ~preview () =
     ] in
   let trees = Array.to_list @@ Sys.readdir tree_dir in
   let dataset_l = derive ~tree_dir ~trees ~preview in
-  let repo = repo_of_dataset_l dataset_l in
+  let repo = repo_of_dataset_l dataset_l ~preview in
   Repo.build ~outdir ~np ~mem:(`GB mem) ~logger repo
 
 let command =