From 57d82513b25d75752ad17b8cc5606add01699f09 Mon Sep 17 00:00:00 2001
From: Floreal Cabanettes <floreal.cabanettes@inra.fr>
Date: Thu, 29 Mar 2018 16:59:37 +0200
Subject: [PATCH] Refactoring
---
build_pop.py | 4 +-
exceptions.py => lib/exceptions.py | 0
.../genotype_results.py | 46 +++++++++++--------
.../variants_simulator.py | 2 +-
4 files changed, 30 insertions(+), 22 deletions(-)
rename exceptions.py => lib/exceptions.py (100%)
rename genotype_results.py => lib/genotype_results.py (93%)
rename variants_simulator.py => lib/variants_simulator.py (99%)
diff --git a/build_pop.py b/build_pop.py
index c12f436..f5acfbb 100755
--- a/build_pop.py
+++ b/build_pop.py
@@ -11,8 +11,8 @@ import vcf
from Bio import SeqIO
import tempfile
from pysam import tabix_compress, tabix_index, VariantFile
-from variants_simulator import VariantsSimulator
-from exceptions import InputException, ExecException
+from lib.variants_simulator import VariantsSimulator
+from lib.exceptions import InputException, ExecException
import multiprocessing
import subprocess
diff --git a/exceptions.py b/lib/exceptions.py
similarity index 100%
rename from exceptions.py
rename to lib/exceptions.py
diff --git a/genotype_results.py b/lib/genotype_results.py
similarity index 93%
rename from genotype_results.py
rename to lib/genotype_results.py
index a236c05..2199f1f 100755
--- a/genotype_results.py
+++ b/lib/genotype_results.py
@@ -65,25 +65,7 @@ def getvarsize(variant):
return variant.stop - variant.start + 1
-def main(genotypes, predicted, filtered, output, type_v="del", verbose=False):
- true_genotypes = readgenotypes(vcffile=genotypes,
- variants={},
- type_v=type_v)
- pred_genotypes = {}
- filtered_genotypes = {}
- with open(predicted, "r") as preds:
- for pred in preds:
- pred = pred.rstrip()
- if pred != "":
- readgenotypes(vcffile=pred,
- variants=pred_genotypes,
- type_v=type_v)
- if filtered:
- readgenotypes(vcffile=pred,
- variants=filtered_genotypes,
- type_v=type_v,
- dofilter=True)
-
+def build(true_genotypes, pred_genotypes, filtered_genotypes, output):
true_pybed = variants_to_pybed(true_genotypes)
pred_pybed = variants_to_pybed(pred_genotypes)
@@ -103,6 +85,32 @@ def main(genotypes, predicted, filtered, output, type_v="del", verbose=False):
(sv[3], "pass", varsize, correct / (wrong + correct), left_prec, right_prec))
out.write("%s\t%s\t%d\t%3.2f\t%d\t%d\n" %
(sv[3], soft, varsize, correct / (wrong + correct), left_prec, right_prec))
+ return pred_with_true
+
+
+def main(genotypes, predicted, filtered, output, type_v="del", verbose=False):
+ true_genotypes = readgenotypes(vcffile=genotypes,
+ variants={},
+ type_v=type_v)
+ pred_genotypes = {}
+ filtered_genotypes = {}
+ with open(predicted, "r") as preds:
+ for pred in preds:
+ pred = pred.rstrip()
+ if pred != "":
+ readgenotypes(vcffile=pred,
+ variants=pred_genotypes,
+ type_v=type_v)
+ if filtered:
+ readgenotypes(vcffile=pred,
+ variants=filtered_genotypes,
+ type_v=type_v,
+ dofilter=True)
+
+ pred_with_true = build(true_genotypes=true_genotypes,
+ pred_genotypes=pred_genotypes,
+ filtered_genotypes=filtered_genotypes,
+ output=output)
if verbose:
print("%d variants" % (len(true_genotypes)))
diff --git a/variants_simulator.py b/lib/variants_simulator.py
similarity index 99%
rename from variants_simulator.py
rename to lib/variants_simulator.py
index d05d551..8d5fc4f 100755
--- a/variants_simulator.py
+++ b/lib/variants_simulator.py
@@ -5,7 +5,7 @@ import sys
import re
import random
-from exceptions import InputException
+from lib.exceptions import InputException
from Bio import SeqIO
from joblib import Parallel, delayed
import multiprocessing
--
GitLab