refactor output directory as "datasets"

lib/pipeline.ml

@@ -22,14 +22,42 @@ let select_out parsed_tree t = match t with
     | Tree4simu    -> parsed_tree / selector [ "tree.only_node_ids.nhx" ]
     | Tree_diffsel -> parsed_tree / selector [ "tree.diffsel" ]
 
+let ready_dataset_of_raw_dataset raw_dataset =
+    let tree = raw_dataset.input_tree in
+    let fna = raw_dataset.fna in
+    let parsed_tree = parse_input_tree ~tree in
+    let faa = Bppsuite.fna2faa ~fna in
+    { parsed_tree; fna; faa}
+
+let repo_of_ready_dataset ready_dataset =
+  Repo.[
+    item ["ready_dataset/tree.H0.node_ids" ] (ready_dataset.parsed_tree / selector [ "tree.H0.node_ids" ]) ;
+    item ["ready_dataset/tree.Ha.node_ids" ] (ready_dataset.parsed_tree / selector [ "tree.Ha.node_ids" ]) ;
+    item ["ready_dataset/tree.only_convergent_tags.nhx" ] (ready_dataset.parsed_tree / selector [ "tree.only_convergent_tags.nhx" ]) ;
+    item ["ready_dataset/tree.only_node_ids.nhx" ] (ready_dataset.parsed_tree / selector [ "tree.only_node_ids.nhx" ]) ;
+    item ["ready_dataset/tree.diffsel" ] (ready_dataset.parsed_tree / selector [ "tree.diffsel" ]) ;
+    item ["ready_dataset/simulated_sequences.fna"] ready_dataset.fna ;
+    item ["ready_dataset/simulated_sequences.faa"] ready_dataset.faa ;
+  ]
+
+let repo_of_raw_dataset raw_dataset =
+  Repo.[
+    item ["raw_dataset/input_tree.nhx"] raw_dataset.input_tree ;
+    item ["raw_dataset/simulated_sequences.fna"] raw_dataset.fna ;
+  ]
+
 let derive_from_tree ~tree_dir ~tree ~preview =
   let tree = input (Filename.concat tree_dir tree) in
   let nb_sites = if preview then 10 else 100 in
   let parsed_tree = parse_input_tree ~tree in
   let fna = Bppsuite.bppseqgen ~nb_sites ~tree:(select_out parsed_tree Tree4simu) in
-  Repo.[
-    item ["simulated_sequences.fna"] fna ;
-  ]
+  let faa = Bppsuite.fna2faa ~fna in
+  let raw_dataset = { input_tree = tree ; fna} in
+  let ready_dataset = { parsed_tree; fna; faa} in
+  if preview then
+    repo_of_ready_dataset ready_dataset
+  else
+    repo_of_raw_dataset raw_dataset
 
 let derive ~tree_dir ~trees ~preview =
   List.map trees ~f:(fun tree ->