From eb62440103fee2283569dd637b39f72781ff3a59 Mon Sep 17 00:00:00 2001
From: Floreal Cabanettes <floreal.cabanettes@inra.fr>
Date: Fri, 23 Mar 2018 14:52:11 +0100
Subject: [PATCH] Fix for pysam 0.14: sort vcf file
---
build_pop.py | 16 +++++++++++-----
1 file changed, 11 insertions(+), 5 deletions(-)
diff --git a/build_pop.py b/build_pop.py
index 3647095..1ea4a0b 100755
--- a/build_pop.py
+++ b/build_pop.py
@@ -189,8 +189,7 @@ def build_genotypes_vcf_list(deletions: dict, inversions: dict, output_vcf, hapl
# Build VCF header:
_build_vcf_header(vcf_file, prg_path, tmp_dir, nb_inds)
- vcf_reader = vcf.Reader(filename=vcf_file)
- vcf_writer = vcf.Writer(open(output_vcf, "w"), vcf_reader)
+ records = []
# Deletions:
for chrm, deletes in deletions.items():
@@ -213,7 +212,7 @@ def build_genotypes_vcf_list(deletions: dict, inversions: dict, output_vcf, hapl
info = {"END": delete["end"], "AF": delete["freq"]}
vcf_record = vcf.model._Record(chrm, delete["start"], delete["name"], "N", [vcf.model._SV("DEL")], ".",
".", info, "GT", [0], genotypes)
- vcf_writer.write_record(vcf_record)
+ records.append(vcf_record)
# Inversions:
genotypes_for_inds_INV = OrderedDict()
@@ -237,9 +236,16 @@ def build_genotypes_vcf_list(deletions: dict, inversions: dict, output_vcf, hapl
info = {"END": inversion["end"], "AF": inversion["freq"]}
vcf_record = vcf.model._Record(chrm, inversion["start"], inversion["name"], "N", [vcf.model._SV("INV")],
".", ".", info, "GT", [0], genotypes)
- vcf_writer.write_record(vcf_record)
+ records.append(vcf_record)
- vcf_writer.close()
+ records.sort(key=lambda r: (r.CHROM, r.start))
+
+ with open(output_vcf, "w") as o_vcf:
+ vcf_reader = vcf.Reader(filename=vcf_file)
+ vcf_writer = vcf.Writer(o_vcf, vcf_reader)
+ for record in records:
+ vcf_writer.write_record(record)
+ vcf_writer.close()
tabix_compress(output_vcf, output_vcf + ".gz", True)
tabix_index(output_vcf + ".gz", force=True, preset="vcf")
--
GitLab