def testBaseCountsTwoReads(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['base_counts'], {
'a': 4,
'c': 5,
't': 4,
'g': 4
})
def testBaseCountsOneRead(self):
seq = '>hey\nagtcagtcagtc'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['base_counts'], {
'a': 3,
'c': 3,
't': 3,
'g': 3
})
def testMedianThreeStrings(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg\n>how\natgggtc'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['median_length'], 7)
def testMedianOneString(self):
seq = '>hey\nagtcagtcagtc'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['median_length'], 12)
def testMedianEmptyInput(self):
result = summarizeReads(StringIO(), 'fasta')
self.assertEqual(result['median_length'], 0)
def testMinLengthListTwoStrings(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['min_length'], 5)
def testMinLengthListEmptyInput(self):
result = summarizeReads(StringIO(), 'fasta')
self.assertEqual(result['min_length'], 0)
def testBaseCountsTwoReads(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['base_counts'],
{'a': 4, 'c': 5, 't': 4, 'g': 4})
def testBaseCountsEmptyImput(self):
result = summarizeReads(StringIO(), 'fasta')
self.assertEqual(result['base_counts'], {})
def testTotalLengthTwoStrings(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['total_length'], 17)
def testTotalLengthOneString(self):
seq = '>hey\nagtcagtcagtc'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['total_length'], 12)
def testTotalLengthEmptyInput(self):
result = summarizeReads(StringIO(), 'fasta')
self.assertEqual(result['total_length'], 0)
def testReadNumberTwoSequencesCount(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['read_number'], 2)
def testReadNumberOneSequenceCount(self):
seq = '>hey\nagtcagtcagtc'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['read_number'], 1)
def testReadNumberEmptyInput(self):
result = summarizeReads(StringIO(), 'fasta')
self.assertEqual(result['read_number'], 0)
def testMedianFourStrings(self):
seq = '>hey\nagtcagtcagtc\n>you\nacctg\n>how\natgggtc\n>are\n\
atggctattgaactgtatct'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['median_length'], 9.5)
from __future__ import print_function
from dark.summarize import summarizeReads
import sys
if len(sys.argv) != 2:
print("Usage: %s file.fasta / file.fastq" % sys.argv[0], file=sys.stderr)
sys.exit(1)
else:
filename = sys.argv[1]
if filename.lower().endswith("q"):
filetype = "fastq"
else:
# Assume anything not ending with a 'q' is fasta.
filetype = "fasta"
result = summarizeReads(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 in sorted(result["base_counts"]):
count = result["base_counts"][base]
print("%s: Total: %s; Average per read: %s" % (base, count, count / result["read_number"]))
def testBaseCountsOneRead(self):
seq = '>hey\nagtcagtcagtc'
result = summarizeReads(StringIO(seq), 'fasta')
self.assertEqual(result['base_counts'],
{'a': 3, 'c': 3, 't': 3, 'g': 3})
