def test_report_barcodes_for_no_counts(self):
"""report_barcodes: check output when there are no counts
"""
bc = BarcodeCounter()
analysis = bc.analyse()
reporter = report_barcodes(bc)
# Check content
self.assertEqual(str(reporter),
"""Barcode analysis for all lanes
==============================
No barcodes counted""")
def test_report_barcodes_for_no_counts(self):
"""report_barcodes: check output when there are no counts
"""
bc = BarcodeCounter()
analysis = bc.analyse()
reporter = report_barcodes(bc)
# Check content
self.assertEqual(str(reporter),
"""Barcode analysis for all lanes
==============================
No barcodes counted""")
def test_report_barcodes(self):
"""report_barcodes: check output for mismatches and sample sheet
"""
# Create sample sheet
sample_sheet_file = self._make_file("SampleSheet.csv",
"""[Data]
Lane,Sample_ID,Sample_Name,Sample_Plate,Sample_Well,I7_Index_ID,index,Sample_Project,Description
1,SMPL1,,,,A006,CATGCGCGGTA,,
1,SMPL2,,,,A012,GCTGCGCGGTC,,
""")
# Set up barcode counts
bc = BarcodeCounter()
bc.count_barcode("TATGCGCGGTA",lane=1,incr=285302)
bc.count_barcode("CATGCGCGGTA",lane=1,incr=8532)
bc.count_barcode("GATGCGCGGTA",lane=1,incr=5321)
bc.count_barcode("GCTGCGCGGTA",lane=1,incr=7853)
bc.count_barcode("GCTGCGCGGTC",lane=1,incr=325394)
analysis = bc.analyse(lane=1,mismatches=2,
sample_sheet=sample_sheet_file)
##"CATGCGCGGTA","TATGCGCGGTA","GATGCGCGGTA","GCTGCGCGGTA" = 307008
##"GCTGCGCGGTC" = 325394
self.assertEqual(analysis.cutoff,None)
self.assertEqual(analysis.mismatches,2)
self.assertEqual(analysis.total_reads,632402)
self.assertEqual(analysis.coverage,632402)
self.assertEqual(analysis.barcodes,["GCTGCGCGGTC",
"CATGCGCGGTA"])
self.assertEqual(analysis.counts["GCTGCGCGGTC"].reads,325394)
self.assertEqual(analysis.counts["CATGCGCGGTA"].reads,307008)
self.assertEqual(analysis.counts["GCTGCGCGGTC"].sample,"SMPL2")
self.assertEqual(analysis.counts["CATGCGCGGTA"].sample,"SMPL1")
self.assertEqual(analysis.counts["GCTGCGCGGTC"].sequences,1)
self.assertEqual(analysis.counts["CATGCGCGGTA"].sequences,4)
# Create report
reporter = report_barcodes(bc,
lane=1,
mismatches=2,
sample_sheet=sample_sheet_file)
# Check content
self.assertEqual(str(reporter),
"""Barcode analysis for lane #1
============================
Barcodes have been grouped by allowing 2 mismatches
#Rank Index Sample N_seqs N_reads %reads (%Total_reads)
1 GCTGCGCGGTC SMPL2 1 325394 51.5% (51.5%)
2 CATGCGCGGTA SMPL1 4 307008 48.5% (100.0%)""")
def test_report_barcodes(self):
"""report_barcodes: check output for mismatches and sample sheet
"""
# Create sample sheet
sample_sheet_file = self._make_file("SampleSheet.csv",
"""[Data]
Lane,Sample_ID,Sample_Name,Sample_Plate,Sample_Well,I7_Index_ID,index,Sample_Project,Description
1,SMPL1,,,,A006,CATGCGCGGTA,,
1,SMPL2,,,,A012,GCTGCGCGGTC,,
""")
# Set up barcode counts
bc = BarcodeCounter()
bc.count_barcode("TATGCGCGGTA",lane=1,incr=285302)
bc.count_barcode("CATGCGCGGTA",lane=1,incr=8532)
bc.count_barcode("GATGCGCGGTA",lane=1,incr=5321)
bc.count_barcode("GCTGCGCGGTA",lane=1,incr=7853)
bc.count_barcode("GCTGCGCGGTC",lane=1,incr=325394)
analysis = bc.analyse(lane=1,mismatches=2,
sample_sheet=sample_sheet_file)
##"CATGCGCGGTA","TATGCGCGGTA","GATGCGCGGTA","GCTGCGCGGTA" = 307008
##"GCTGCGCGGTC" = 325394
self.assertEqual(analysis.cutoff,None)
self.assertEqual(analysis.mismatches,2)
self.assertEqual(analysis.total_reads,632402)
self.assertEqual(analysis.coverage,632402)
self.assertEqual(analysis.barcodes,["GCTGCGCGGTC",
"CATGCGCGGTA"])
self.assertEqual(analysis.counts["GCTGCGCGGTC"].reads,325394)
self.assertEqual(analysis.counts["CATGCGCGGTA"].reads,307008)
self.assertEqual(analysis.counts["GCTGCGCGGTC"].sample,"SMPL2")
self.assertEqual(analysis.counts["CATGCGCGGTA"].sample,"SMPL1")
self.assertEqual(analysis.counts["GCTGCGCGGTC"].sequences,1)
self.assertEqual(analysis.counts["CATGCGCGGTA"].sequences,4)
# Create report
reporter = report_barcodes(bc,
lane=1,
mismatches=2,
sample_sheet=sample_sheet_file)
# Check content
self.assertEqual(str(reporter),
"""Barcode analysis for lane #1
============================
Barcodes have been grouped by allowing 2 mismatches
#Rank Index Sample N_seqs N_reads %reads (%Total_reads)
1 GCTGCGCGGTC SMPL2 1 325394 51.5% (51.5%)
2 CATGCGCGGTA SMPL1 4 307008 48.5% (100.0%)""")
counts = count_barcodes_bcl2fastq(args[0])
else:
# Generate counts from fastq files
counts = count_barcodes(args)
# Deal with cutoff
if opts.cutoff == 0.0:
cutoff = None
else:
cutoff = opts.cutoff
# Report the counts
if not opts.no_report:
reporter = Reporter()
if lanes is None:
report_barcodes(counts,
cutoff=cutoff,
sample_sheet=opts.sample_sheet,
mismatches=opts.mismatches,
reporter=reporter)
else:
for lane in lanes:
if lane not in counts.lanes:
logging.error("Requested analysis for lane %d but "
"only have counts for lanes %s" %
(lane,
','.join([str(l) for l in counts.lanes])))
sys.exit(1)
report_barcodes(counts,
lane=lane,
cutoff=cutoff,
sample_sheet=opts.sample_sheet,
mismatches=opts.mismatches,
counts = count_barcodes_bcl2fastq(args[0])
else:
# Generate counts from fastq files
counts = count_barcodes(args)
# Deal with cutoff
if opts.cutoff == 0.0:
cutoff = None
else:
cutoff = opts.cutoff
# Report the counts
if not opts.no_report:
reporter = Reporter()
if lanes is None:
report_barcodes(counts,
cutoff=cutoff,
sample_sheet=opts.sample_sheet,
mismatches=opts.mismatches,
reporter=reporter)
else:
for lane in lanes:
if lane not in counts.lanes:
logging.error("Requested analysis for lane %d but "
"only have counts for lanes %s" %
(lane,
','.join([str(l) for l in counts.lanes])))
sys.exit(1)
report_barcodes(counts,
lane=lane,
cutoff=cutoff,
sample_sheet=opts.sample_sheet,
mismatches=opts.mismatches,
