Allow to use custom frequencies for variants

build_pop.py

@@ -69,6 +69,10 @@ def parse_args():
                         const=True, default=False)
     parser.add_argument("-pd", "--proba-del", help="Probabilty to have a deletion", type=float, default=0.000001)
     parser.add_argument("-pi", "--proba-inv", help="Probablity to have an insertion", type=float, default=0.000001)
+    parser.add_argument('-fd', '--freq-del', type=float, required=False,
+                        help="Frequencies choices for deletions, space separated", nargs="+", default=[0.2, 0.5])
+    parser.add_argument('-fi', '--freq-inv', type=float, required=False,
+                        help="Frequencies choices for inversions, space separated", nargs="+", default=[0.2, 0.5])
     parser.add_argument("-l", "--read-len", help="Generate reads having a length of LEN", type=int, default=100)
     parser.add_argument("-m", "--insert-len-mean", help="Generate inserts (fragments) having an average length of LEN",
                         type=int, default=300)
@@ -528,8 +532,8 @@ def confirm(deletions: dict, inversions: dict, variants: dict, printer: Print, a
     return True
 
 
-def init(output_dir, force_outputdir, sv_list, nstretches, nb_inds, reference, proba_del, proba_inv, haploid,
-         force_polymorphism, coverage, read_len, insert_len_mean, insert_len_sd, threads, genotypes=None,
+def init(output_dir, force_outputdir, sv_list, nstretches, nb_inds, reference, proba_del, proba_inv, freq_del, freq_inv,
+         haploid, force_polymorphism, coverage, read_len, insert_len_mean, insert_len_sd, threads, genotypes=None,
          min_deletions=1, min_inversions=1, max_try=10, quiet=True, stdout=None, stderr=None):
 
     printer = Print(stdout=stdout, stderr=stderr)
@@ -576,7 +580,8 @@ def init(output_dir, force_outputdir, sv_list, nstretches, nb_inds, reference, p
             nb_deletions = -1  # To allow 0 deletions
             nb_inversions = -1  # To allow 0 inversions
             nb_try = 0
-            sv_sim = VariantsSimulator(sv_list, nstretches, threads, stdout, stderr)
+            sv_sim = VariantsSimulator(sv_list=sv_list, nstretches=nstretches, threads=threads, stdout=stdout,
+                                       stderr=stderr, freq_del=freq_del, freq_inv=freq_inv)
             if "DEL" not in sv_sim.variants:
                 proba_del = 0
             if "INV" not in sv_sim.variants:

lib/variants_simulator.py

@@ -23,7 +23,8 @@ class bcolors:
 
 
 class VariantsSimulator:
-    def __init__(self, sv_list: str = None, nstretches: str = None, threads: int = -1, stdout=None, stderr=None):
+    def __init__(self, sv_list: str = None, freq_del: list=(0.2, 0.5), freq_inv: list=(0.2, 0.5), nstretches: str = None,
+                 threads: int = -1, stdout=None, stderr=None):
         self._epsilon = None
         if sv_list is None:
             sv_list = os.path.join(os.path.dirname(os.path.dirname(os.path.realpath(__file__))), "defaults.rules")
@@ -34,6 +35,8 @@ class VariantsSimulator:
             self.nstretches = self.__load_nstretches(nstretches)
         self.deletions = {}
         self.invertions = {}
+        self.freq_del = freq_del
+        self.freq_inv = freq_inv
         self.threads = threads if threads > -1 else multiprocessing.cpu_count()
         self.stdout = stdout
         self.stderr = stderr
@@ -303,7 +306,7 @@ class VariantsSimulator:
                     else:
                         self.print_ok("N-stretch filter: OK", print_message_header)
                     i = start
-                    freq = 0.2 if random.uniform(0, 1) < 0.5 else 0.5
+                    freq = random.choice(self.freq_del)
                     deletions[chrm].append({
                         "name": "DEL{0}_{1}".format(chrm, nb_dels_on_ch),
                         "start": start,
@@ -345,7 +348,7 @@ class VariantsSimulator:
                     else:
                         self.print_ok("N-stretch filter: OK", print_message_header)
                 i = start
-                freq = 0.2 if random.uniform(0, 1) < 0.5 else 0.5
+                freq = random.choice(self.freq_inv)
                 inversions[chrm].append({
                     "name": "INV{0}_{1}".format(chrm, nb_inv_on_ch),
                     "start": start,