started alternative pipeline formulation

lib/pipeline2.ml

+open Base
+open Printf
+open Bistro.EDSL
+open File_formats
+
+type dataset =
+  | Simulation of simulation
+
+and simulation = {
+  hypothesis : Convergence_hypothesis.t ;
+  tree : [`NHX of string] ;
+  profiles : string ;
+  nb_sites : int ;
+  seed : int ;
+}
+
+let calc_fixed_seed ~(str:string) (seed:int) : int =
+  let str_hash = Hashtbl.hash str in
+  Hashtbl.hash (str_hash + seed)
+
+let simulation ~hyp ~tree ~profiles ~nb_sites ~seed =
+  Simulation {
+    hypothesis = hyp ;
+    tree = `NHX tree ;
+    profiles ;
+    nb_sites ;
+    seed ;
+  }
+
+let tree_prefix = function
+  | Simulation { tree = `NHX path } ->
+    Caml.Filename.chop_extension path
+
+let tree_dataset = function
+  | Simulation { tree = `NHX tree } as sim ->
+    Tree_dataset.prepare
+      ~descr:("simulated_data." ^ (tree_prefix sim))
+      (input tree)
+
+let tree = function
+  | Simulation { tree = `NHX path } -> input path
+
+let profile sim =
+  Profile.profile_l_of_splitted_profile
+    ~nb_cat:All
+    ~nb_sites:sim.nb_sites
+    sim.profiles
+    ~seed:(Random.int Int.max_value)
+
+let bppseqgen (Simulation sim as d) =
+  let model_prefix = Convergence_hypothesis.string_of_model sim.hypothesis in
+  let descr = sprintf ".%s.%s" model_prefix (tree_prefix d) in
+  let profile = profile sim in
+  let profile_f = profile.profile_f in
+  let profile_c = profile.profile_c in
+  let seed = calc_fixed_seed ~str:descr sim.seed in
+  Bppsuite.Bppseqgen.multi_profiles
+    ~descr
+    ~tree_dataset:(tree_dataset d)
+    ~hypothesis:sim.hypothesis ~profile_f ~profile_c ~seed
+
+let nucleotide_alignment (Simulation sim as d) =
+  bppseqgen d
+  |> Bppsuite.Bppseqgen.alignment
+
+let phylip_nucleotide_alignment d =
+  Bppsuite.fna2phy (nucleotide_alignment d)
+
+(* let diffseldsparse d =
+ *   Diffseldsparse.diffseldsparse
+ *     ~alignment:(phylip_nucleotide_alignment d)
+ *     ~tree:(tree d)
+ *     ~w_every:1
+ *     ~n_cycles:2
+ *     () *)
+
+let eval x =
+  Bistro_utils.Term.(
+    run ~np:8 ~mem:(`GB 8) ~logger:(Bistro_utils.Console_logger.create ()) (pureW x)
+  )