def test_gc_content(self):
pytest.raises(seq.NullSequenceError, seq.gc_content, "")
pytest.raises(seq.NullSequenceError, seq.gc_content, "NNNNN")
assert seq.gc_content("AAATTT") == 0.0
assert seq.gc_content("GGGGCC") == 1.0
assert seq.gc_content("ACGTACGT") == 0.5
assert seq.gc_content("ACCTACGT") == 0.5
def test_gc_content(self):
pytest.raises(seq.NullSequenceError, seq.gc_content, "")
pytest.raises(seq.NullSequenceError, seq.gc_content, "NNNNN")
assert seq.gc_content("AAATTT") == 0.0
assert seq.gc_content("GGGGCC") == 1.0
assert seq.gc_content("ACGTACGT") == 0.5
assert seq.gc_content("ACCTACGT") == 0.5
def summary_statistics(reads):
"""Returns a dictionary of summary statistics of the reads. The keys are:
"rnames": a Counter of the rnames of the reads
"flags": a Counter of the bitflags of the reads
"cigars": a Counter of the string representations of the
cigars of the reads
"gc": the average GC content of the sequences.
"read_count": the number of sam reads
"hash": a Counter of the qnames of the reads
"rnames_mapped": a summary of the number of reads with each rname
"""
summary = {
"rnames": Counter(),
"flags": Counter(),
"cigars": Counter(),
"gc": 0,
"read_count": 0,
"hashes": Counter()
}
for read in reads:
summary["rnames"][read.rname] += 1
summary["flags"][read.flag] += 1
summary["cigars"][str(read.cigar)] += 1
summary["gc"] += seq.gc_content(read.seq)
summary["read_count"] += 1
summary["hashes"][read.qname] += 1
summary["gc"] /= summary["read_count"]
return summary
def summary_statistics(reads):
"""Returns a dictionary of summary statistics of the reads. The keys are:
"rnames": a Counter of the rnames of the reads
"flags": a Counter of the bitflags of the reads
"cigars": a Counter of the string representations of the
cigars of the reads
"gc": the average GC content of the sequences.
"read_count": the number of sam reads
"hash": a Counter of the qnames of the reads
"rnames_mapped": a summary of the number of reads with each rname
"""
summary = {"rnames": Counter(),
"flags": Counter(),
"cigars": Counter(),
"gc": 0,
"read_count": 0,
"hashes": Counter()}
for read in reads:
summary["rnames"][read.rname] += 1
summary["flags"][read.flag] += 1
summary["cigars"][str(read.cigar)] += 1
summary["gc"] += seq.gc_content(read.seq)
summary["read_count"] += 1
summary["hashes"][read.qname] += 1
summary["gc"] /= summary["read_count"]
return summary
