added signature for simulation_dataset

app/diffseldsparse_benchmark.ml

@@ -4,14 +4,10 @@ open Reviewphiltrans
 let main ~n_h0 ~n_ha ~seed:i () =
   let open Simulation_dataset in
   let sim =
-    Bppseqgen_mixed {
-      tree = `NHX "example/trees_analyses/cyp_coding.Chrysithr_root.nhx" ;
-      profiles = "example/aa_fitness/263SelectedProfiles.tsv" ;
-      n_h0 ;
-      n_ha ;
-      ne_s = 4. ;
-      seed = i ;
-    }
+    bppseqgen_mixed_simulation
+      ~tree:(NHX "example/trees_analyses/cyp_coding.Chrysithr_root.nhx")
+      ~profiles:"example/aa_fitness/263SelectedProfiles.tsv"
+      ~n_h0 ~n_ha ~ne_s:4. ~seed:i ()
   in
   let w = benchmark sim in
   Bistro_engine.Scheduler.simple_eval_exn ~np:4 ~mem:(`GB 4) (Bistro.Workflow.path w)

lib/simulation_dataset.ml

@@ -6,7 +6,13 @@ let calc_fixed_seed ~(str:string) (seed:int) : int =
   let str_hash = Hashtbl.hash str in
   Hashtbl.hash (str_hash + seed)
 
-type tree = [`NHX of string | `Pair_tree of float * float * int]
+type tree =
+  | NHX of string
+  | Pair_tree of {
+      npairs : int ;
+      branch_length1 : float ;
+      branch_length2 : float ;
+    }
 
 type t =
   | Bppseqgen_simulation of {
@@ -67,10 +73,10 @@ let convdet_simulation ?(branch_factor = 1.) ?(ne_s = 1., 1.) ?(gBGC = 0., 0.) ?
   }
 
 let prefix_of_tree = function
-  | `NHX path ->
+  | NHX path ->
     Caml.Filename.chop_extension path
-  | `Pair_tree (bl1, bl2, np) ->
-    sprintf "pair_tree(bl1=%g,bl2=%g,np=%d)" bl1 bl2 np
+  | Pair_tree { branch_length1 ; branch_length2 ; npairs } ->
+    sprintf "pair_tree(bl1=%g,bl2=%g,np=%d)" branch_length1 branch_length2 npairs
 
 let tree_prefix = function
   | Bppseqgen_simulation { tree ; _ }
@@ -83,18 +89,13 @@ let tree ~branch_length_unit:_ = function
   | Bppseqgen_mixed { tree ; _ }
   | Convdet_simulation { tree ; _ } ->
     match tree with
-    | `NHX path ->
+    | NHX path ->
       Workflow.input path
-    | `Pair_tree (branch_length1,
-                  branch_length2,
-                  npairs) ->
+    | Pair_tree { branch_length1 ;
+                  branch_length2 ;
+                  npairs } ->
     Simulator.pair_tree ~branch_length1 ~branch_length2 ~npairs
 
-let tree_dataset sim =
-  Tree_dataset.prepare
-    ~descr:("simulated_data." ^ (tree_prefix sim))
-    (tree ~branch_length_unit:`Nucleotide sim)
-
 let seed = function
   | Bppseqgen_mixed s -> s.seed
   | Bppseqgen_simulation s -> s.seed
@@ -227,7 +228,7 @@ let explore_params ~(f: param_t -> _) =
 let simu_of_param ?n_h0:(n_h0=50) (p: param_t) =
   let bf, gbgc = p in
   convdet_simulation
-    ~tree:(`NHX "example/trees_analyses/C4AmaranthaceaePolyroot.nhx")
+    ~tree:(NHX "example/trees_analyses/C4AmaranthaceaePolyroot.nhx")
     ~profiles:"example/aa_fitness/263SelectedProfiles.tsv"
     ~branch_factor:bf
     ~gBGC:(match gbgc with

lib/simulation_dataset.mli

+open Bistro
+
+type tree =
+  | NHX of string
+  | Pair_tree of {
+      npairs : int ;
+      branch_length1 : float ;
+      branch_length2 : float ;
+    }
+
+type t
+
+val bppseqgen_mixed_simulation :
+  ?ne_s:float ->
+  ?seed:int ->
+  tree:tree -> profiles:string -> n_h0:int -> n_ha:int -> unit -> t
+
+val bppseqgen_simulation :
+  hyp:Convergence_hypothesis.t ->
+  tree:tree -> profiles:string -> nb_sites:int -> seed:int -> t
+
+val convdet_simulation :
+ ?branch_factor:float ->
+ ?ne_s:float * float ->
+ ?gBGC:float * float ->
+ ?seed:int ->
+ tree:tree -> profiles:string -> n_h0:int -> n_ha:int -> unit -> t
+
+include Detection_pipeline.Query with type t := t
+include Detection_pipeline.S with type query := t
+
+val alignment_plot : t -> svg file
+
+val multinomial_benchmark : t -> pdf file
+
+val benchmark :
+  ?mode:[`fast | `full] ->
+  t ->
+  svg file
+
+val benchmark2 : t -> pdf file
+
+
+(** stuff for gbgc SBME paper *)
+type gBGC_t = Global of float | Convergent of float * float
+type param_t
+val explore_params : f:(param_t -> 'a) -> ((float * gBGC_t) * 'a) list
+val simu_of_param : ?n_h0:int -> param_t -> t
+val filter_results : f:('a -> bool) -> (param_t * 'a) list -> (param_t * 'a) list
+type record_t
+val record_of_simu : t -> record_t workflow
+val realistic_result : record_t -> bool