Require pysam>=0.21.0; fix mpileup call (#335)

ariba/samtools_variants.py

@@ -1,6 +1,7 @@
 import os
 import sys
 import pysam
+import pysam.bcftools
 import pyfastaq
 import vcfcall_ariba
 
@@ -36,13 +37,11 @@ def _make_vcf_and_read_depths_files(self):
 
         tmp_vcf = self.vcf_file + '.tmp'
         with open(tmp_vcf, 'w') as f:
-            print(pysam.mpileup(
-                '-t', 'INFO/AD,INFO/ADF,INFO/ADR',
+            print(pysam.bcftools.mpileup(
+                '-a', 'INFO/AD,INFO/ADF,INFO/ADR',
                 '-L', '99999999',
                 '-A',
                 '-f', self.ref_fa,
-                '-u',
-                '-v',
                 self.bam,
             ), end='', file=f)
 

ariba/tests/data/samtools_variants_make_vcf_and_depths_files.expect.vcf

@@ -1,4 +1,4 @@
-ref1	121	.	C	A,T,<*>	0	.	DP=142;ADF=72,16,18,0;ADR=22,8,1,0;AD=94,24,19,0;I16=72,22,34,9,3752,149792,1660,65086,3948,165816,1300,42424,1488,30666,682,13620;QS=0.748454,0.135248,0.116298,0;VDB=0.419877;SGB=-0.693146;RPB=0.693393;MQB=2.09583e-09;MQSB=0.677623;BQB=0.54095;MQ0F=0	PL	78,0,255,65,197,255,255,255,255,255
-ref1	149	.	AC	ACC	0	.	INDEL;IDV=162;IMF=0.910112;DP=178;ADF=0,96;ADR=0,79;AD=0,175;I16=0,0,96,79,0,0,6916,273882,0,0,5838,208760,0,0,2791,56033;QS=0,1;VDB=0.470182;SGB=-0.693147;MQSB=0.999996;MQ0F=0	PL	255,255,0
-ref1	201	.	C	G,<*>	0	.	DP=190;ADF=0,80,0;ADR=0,94,0;AD=0,174,0;I16=0,0,80,94,0,0,6796,267896,0,0,6554,257056,0,0,2900,58094;QS=0,1,0;VDB=0.437278;SGB=-0.693147;MQSB=0.849709;MQ0F=0	PL	255,255,0,255,255,255
-ref2	170	.	A	T,<*>	0	.	DP=117;ADF=0,52,0;ADR=0,59,0;AD=0,111,0;I16=0,0,52,59,0,0,4390,174306,0,0,4662,195804,0,0,1824,36678;QS=0,1,0;VDB=0.30232;SGB=-0.693147;MQSB=1;MQ0F=0	PL	255,255,0,255,255,255
+ref1	121	.	C	A,T,<*>	0	.	DP=142;ADF=72,16,20,0;ADR=22,8,2,0;AD=94,24,22,0;I16=72,22,36,10,3752,149792,1684,65278,3948,165816,1408,46528,1488,30666,685,13623;QS=0.744888,0.134604,0.120508,0;VDB=0.740082;SGB=-0.693147;RPBZ=-0.623486;MQBZ=-8.79322;MQSBZ=-1.56952;BQBZ=-3.61356;SCBZ=0;MQ0F=0	PL	77,0,255,51,195,255,255,255,255,255
+ref1	149	.	AC	ACC	0	.	INDEL;IDV=162;IMF=0.910112;DP=178;ADF=6,91;ADR=2,71;AD=8,162;I16=6,2,91,71,320,12800,6480,259200,296,11478,5382,191916,12,22,2780,56012;QS=0.0254111,0.974589;VDB=0.0591298;SGB=-0.693147;RPBZ=1.64808;MQBZ=0.803052;MQSBZ=-1.56952;BQBZ=-3.61356;SCBZ=0;MQ0F=0	PL	255,255,0
+ref1	201	.	C	G,<*>	0	.	DP=190;ADF=0,84,0;ADR=0,99,0;AD=0,183,0;I16=0,0,84,99,0,0,6868,268472,0,0,6842,266992,0,0,2911,58109;QS=0,1,0;VDB=0.708247;SGB=-0.693147;MQSBZ=0.614759;MQ0F=0	PL	255,255,0,255,255,255
+ref2	170	.	A	T,<*>	0	.	DP=117;ADF=0,55,0;ADR=0,62,0;AD=0,117,0;I16=0,0,55,62,0,0,4680,187200,0,0,4914,206388,0,0,1827,36681;QS=0,1,0;VDB=0.655552;SGB=-0.693147;MQSBZ=0;MQ0F=0	PL	255,255,0,255,255,255

ariba/tests/samtools_variants_test.py

@@ -48,7 +48,9 @@ def get_vcf_call_lines(fname):
             got = got_lines[i].split('\t')
             self.assertEqual(len(expected), len(got))
             self.assertEqual(expected[:7], got[:7])
-            self.assertEqual(expected[-2:], got[-2:])
+            # last two cols are PL and then values. These are not used, and
+            # can change slightly between samtools/bcftools versions. Ignore.
+            #self.assertEqual(expected[-2:], got[-2:])
             exp_set = set(expected[7].split(';'))
             got_set = set(got[7].split(';'))
             self.assertEqual(exp_set, got_set)

install_dependencies.sh

@@ -24,7 +24,6 @@ apt-get install --no-install-recommends -y \
   python3-dev \
   python3-setuptools \
   python3-pip \
-  python3-pysam \
   python3-tk \
   python3-matplotlib \
   unzip \
@@ -38,6 +37,8 @@ if [ ! -d $install_root ]; then
 fi
 cd $install_root
 
+# pysam from pip because apt-get version is too old
+python3 -m pip install pysam>=0.21.0
 
 # --------------- bowtie2 ------------------
 cd $install_root

requirements.txt

@@ -2,6 +2,6 @@ BeautifulSoup4 >= 4.1.0
 biopython < 1.78
 dendropy >= 4.2.0
 pyfastaq >= 3.12.0
-pysam >= 0.9.1
+pysam >= 0.21.0
 pymummer <= 0.10.3
 matplotlib >= 3.1.0