From bbf963fb148ddc4648aeb00f0d9f5a7fc3f9e011 Mon Sep 17 00:00:00 2001
From: John Huddleston <huddlej@gmail.com>
Date: Thu, 18 Mar 2021 14:41:45 -0700
Subject: [PATCH] Replace utils.open_file in mask with io.open_file

Replaces calls to a similar `open_file` function from `utils.py` with
the new function in `io.py`. Updates the functional tests for the mask
module to confirm that compressed VCF inputs work with the old and new
function alike.
---
 augur/mask.py                         |   4 ++--
 tests/functional/mask.t               |  14 +++++++-------
 tests/functional/mask/variants.vcf    |  10 ----------
 tests/functional/mask/variants.vcf.gz | Bin 0 -> 155 bytes
 4 files changed, 9 insertions(+), 19 deletions(-)
 delete mode 100644 tests/functional/mask/variants.vcf
 create mode 100644 tests/functional/mask/variants.vcf.gz

diff --git a/augur/mask.py b/augur/mask.py
index f45c5ba52..b52232d7d 100644
--- a/augur/mask.py
+++ b/augur/mask.py
@@ -10,8 +10,8 @@
 from Bio import SeqIO
 from Bio.Seq import MutableSeq
 
-from .io import read_sequences, write_sequences
-from .utils import run_shell_command, shquote, open_file, is_vcf, load_mask_sites, VALID_NUCLEOTIDES
+from .io import open_file, read_sequences, write_sequences
+from .utils import run_shell_command, shquote, is_vcf, load_mask_sites, VALID_NUCLEOTIDES
 
 def get_chrom_name(vcf_file):
     """Read the CHROM field from the first non-header line of a vcf file.
diff --git a/tests/functional/mask.t b/tests/functional/mask.t
index b669e134c..9197b28a3 100644
--- a/tests/functional/mask.t
+++ b/tests/functional/mask.t
@@ -5,24 +5,24 @@ Integration tests for augur mask.
 
 Try masking a VCF without any specified mask.
 
-  $ ${AUGUR} mask --sequences mask/variants.vcf
+  $ ${AUGUR} mask --sequences mask/variants.vcf.gz
   No masking sites provided. Must include one of --mask, --mask-from-beginning, --mask-from-end, --mask-invalid, or --mask-sites
   [1]
 
 Mask a VCF with a BED file and no specified output file.
 
-  $ cp "mask/variants.vcf" "$TMP/"
+  $ cp "mask/variants.vcf.gz" "$TMP/"
   $ ${AUGUR} mask \
-  >  --sequences "$TMP/variants.vcf" \
+  >  --sequences "$TMP/variants.vcf.gz" \
   >  --mask "mask/mask_variants.bed" > /dev/null
 
-  $ diff -u "mask/masked_variants.vcf" "$TMP/variants.vcf"
-  $ rm -f "$TMP/variants.vcf"
+  $ diff -u "mask/masked_variants.vcf" <(gzip -c -d "$TMP/variants.vcf.gz")
+  $ rm -f "$TMP/variants.vcf.gz"
 
 Mask a VCF with a BED file and a specified output file.
 
   $ ${AUGUR} mask \
-  >  --sequences "mask/variants.vcf" \
+  >  --sequences "mask/variants.vcf.gz" \
   >  --mask "mask/mask_variants.bed" \
   >  --output "$TMP/masked_variants.vcf" > /dev/null
 
@@ -104,4 +104,4 @@ Mask invalid nucleotides
   ATCGNNNN
   $ rm -f "$TMP/masked.fasta"
 
-  $ popd > /dev/null
\ No newline at end of file
+  $ popd > /dev/null
diff --git a/tests/functional/mask/variants.vcf b/tests/functional/mask/variants.vcf
deleted file mode 100644
index 06293fd86..000000000
--- a/tests/functional/mask/variants.vcf
+++ /dev/null
@@ -1,10 +0,0 @@
-#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO
-MTB_anc	1199	.	G	A	.	PASS	.
-MTB_anc	1463	.	G	C	.	PASS	.
-MTB_anc	2018	.	C	T	.	PASS	.
-MTB_anc	2532	.	C	T	.	PASS	.
-MTB_anc	3132	.	A	C	.	PASS	.
-MTB_anc	4663	.	G	C	.	PASS	.
-MTB_anc	5887	.	C	T	.	PASS	.
-MTB_anc	7581	.	G	T	.	PASS	.
-MTB_anc	8322	.	T	G	.	PASS	.
diff --git a/tests/functional/mask/variants.vcf.gz b/tests/functional/mask/variants.vcf.gz
new file mode 100644
index 0000000000000000000000000000000000000000..61e4698844bd4196c2b9c9cbe5825c0b5e5b208b
GIT binary patch
literal 155
zcmV;M0A&9kiwFoyP>x>!19o9@X<=@3b1rscW&m}~u@b@{5CqVA{SEt!KuAc57?c?z
z5iYgk*!%ybNTG$$Tybx9cW*8+ELcN=J0Z>%+7oV@_OQ;=9O14u>=ruSAFn4En<jKP
zL+88dl+gXGA4D<pZ~VYi{u)rlq~I+RteCEK_CtG5DmAt`N2yrJRZI!~*`usqdjqky
Ju#zJI003E|Mg{-?

literal 0
HcmV?d00001