add phyml to re infer tree after simu

etc/docker/phyml/Dockerfile

+# PhyML
+# http://www.atgc-montpellier.fr/phyml/
+# v3.1
+FROM debian:stretch
+
+RUN apt-get update && apt-get install -y \
+     wget \
+     unzip
+
+WORKDIR /phyml
+RUN wget http://www.atgc-montpellier.fr/download/binaries/phyml/PhyML-3.1.zip \
+    && unzip PhyML-3.1.zip \
+    && rm -rf PhyML-3.1.zip \
+    && mv PhyML-3.1/PhyML-3.1_linux64 /usr/local/bin/phyml \
+    && chmod 755 /usr/local/bin/phyml \
+    && rm -rf /phyml/PhyML-3.1
+
+WORKDIR /data

etc/docker/phyml/build_docker.sh

+#! /bin/bash
+
+set -e
+
+IMAGE_NAME=phyml
+DOCKERFILE_DIR=.
+TAG=v3.1
+REPO=carinerey/$IMAGE_NAME:$TAG
+docker build -t $REPO -f ./Dockerfile $DOCKERFILE_DIR
+
+if [[ $1 == "push_yes" ]]
+then
+    docker push $REPO
+fi

lib/phyml.ml

+open Core
+open Bistro.Std
+open Bistro.EDSL
+open Bistro_bioinfo.Std
+
+type phyml
+
+let phyml ~tree phy : phyml directory workflow =
+  let env = docker_image ~account:"carinerey" ~name:"phyml" ~tag:"v3.1" () in
+  (*mpirun -np 4 phyml -o lr -u cyp_coding.ok.nwk -i cyp_coding.phy -d 'nt' -m gtr*)
+  let tmp_phy = dest // "tree.phy" in
+  workflow ~descr:"phyml" [
+    docker env (
+      and_list [
+        mkdir_p dest;
+        mkdir_p tmp;
+        cd tmp;
+        cmd "cp" [ dep phy ; ident tmp_phy] ;
+        cmd "phyml" ~env [
+          string "--quiet" ;
+          opt "-u" dep tree;
+          opt "-o" string "lr";
+          opt "-i"  ident tmp_phy ;
+          opt "-d"  string "nt" ;
+          opt "-m"  string "gtr" ;
+        ];
+      ])
+  ]
+
+let phyml_tree ~tree phy =
+  (phyml ~tree phy) / selector ["tree.phy_phyml_tree.txt"]

lib/ready_dataset.ml

@@ -22,10 +22,13 @@ let of_raw (raw_dataset : Raw_dataset.t) =
   let faa = Bppsuite.fna2faa ~fna in
   { input_tree; tree_dataset ; fna; faa; fna_infos}
 
+
 let repo rd =
+  let phy = (Bppsuite.fa2phy rd.fna) in
   Repo.[
     [
       item ["input_tree.nhx"] rd.input_tree ;
+      item ["recalculated_tree.nw"] (Phyml.phyml_tree ~tree:rd.input_tree phy );
       item ["tree.H0.node_ids" ] (Tree_dataset.nodes rd.tree_dataset H0) ;
       item ["tree.Ha.node_ids" ] (Tree_dataset.nodes rd.tree_dataset HaPCOC) ;
       item ["tree.only_convergent_tags.nhx" ] (Tree_dataset.tree rd.tree_dataset `Detection) ;
@@ -33,7 +36,7 @@ let repo rd =
       item ["tree.diffsel" ] (Tree_dataset.diffsel_tree rd.tree_dataset) ;
       item ["tree.convergent_topology" ] (Tree_dataset.topological_tree rd.tree_dataset) ;
       item ["simulated_sequences.fna"] rd.fna ;
-      item ["simulated_sequences.phy"] (Bppsuite.fa2phy rd.fna) ;
+      item ["simulated_sequences.phy"] phy ;
       item ["simulated_sequences.faa"] rd.faa ;
     ] ;
     match rd.fna_infos with