def testMedianFourStrings(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg\n>how\natgggtc\n>are\n\
atggctattgaactgtatct'
result = summarize_reads(StringIO(seq), 'fasta')
self.assertEqual(result['median_length'], 9.5)
def testMedianOneString(self):
seq = '>hey\nagtcagtcagtc'
result = summarize_reads(StringIO(seq), 'fasta')
self.assertEqual(result['median_length'], 12)
def testMedianThreeStrings(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg\n>how\natgggtc'
result = summarize_reads(StringIO(seq), 'fasta')
self.assertEqual(result['median_length'], 7)
def testMinLengthListTwoStrings(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg'
result = summarize_reads(StringIO(seq), 'fasta')
self.assertEqual(result['min_length'], 5)
def testMedianEmptyInput(self):
result = summarize_reads(StringIO(), 'fasta')
self.assertEqual(result['median_length'], 0)
def testBaseCountsTwoReads(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg'
result = summarize_reads(StringIO(seq), 'fasta')
self.assertEqual(result['base_counts'],
{'a': 4, 'c': 5, 't': 4, 'g': 4})
def testMinLengthListEmptyInput(self):
result = summarize_reads(StringIO(), 'fasta')
self.assertEqual(result['min_length'], 0)
def testBaseCountsEmptyImput(self):
result = summarize_reads(StringIO(), 'fasta')
self.assertEqual(result['base_counts'], {})
def testBaseCountsOneRead(self):
seq = '>hey\nagtcagtcagtc'
result = summarize_reads(StringIO(seq), 'fasta')
self.assertEqual(result['base_counts'],
{'a': 3, 'c': 3, 't': 3, 'g': 3})
def testTotalLengthOneString(self):
seq = '>hey\nagtcagtcagtc'
result = summarize_reads(StringIO(seq), 'fasta')
self.assertEqual(result['total_length'], 12)
def testTotalLengthTwoStrings(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg'
result = summarize_reads(StringIO(seq), 'fasta')
self.assertEqual(result['total_length'], 17)
def testTotalLengthEmptyInput(self):
result = summarize_reads(StringIO(), 'fasta')
self.assertEqual(result['total_length'], 0)
def testReadNumberTwoSequencesCount(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg'
result = summarize_reads(StringIO(seq), 'fasta')
self.assertEqual(result['read_number'], 2)
def testReadNumberOneSequenceCount(self):
seq = '>hey\nagtcagtcagtc'
result = summarize_reads(StringIO(seq), 'fasta')
self.assertEqual(result['read_number'], 1)
def testReadNumberEmptyInput(self):
result = summarize_reads(StringIO(), 'fasta')
self.assertEqual(result['read_number'], 0)
#!/usr/bin/env python
from dark.summarize import summarize_reads
import sys
if len(sys.argv) > 2:
print >>sys.stderr, "Usage: %s file.fasta / file.fastq" % sys.argv[0]
sys.exit(1)
else:
filename = sys.argv[1]
if filename.endswith('a'):
filetype = 'fasta'
else:
filetype = 'fastq'
result = summarize_reads(filename, filetype)
print "Number of reads:", result["read_number"]
print "Total length: %s bases" % result["total_length"]
print "The average read length: %s bases" % result["average_length"]
print "Longest read:", result["max_length"]
print "Shortest read:", result["min_length"]
print "Median length:", result["median_length"]
for base, count in result["base_counts"].items():
print "%s: Total: %s; Average per read: %s" % (
base, count, count / result["read_number"])