def test_optionalthird(self):
''' Test that given a correct third argument it still runs '''
infa = ungzip(INPUT_PATH)
bwa.index_ref(REF_PATH)
# I don't have a mates file so just use infa again
mem = BWAMem(REF_PATH, infa, infa, bwa_path=BWA_PATH)
eq_(0, mem.run())
def test_optionspassed(self):
''' Make sure all options passed make it to bwa '''
bwa = self.mkbwa('$@')
mem = BWAMem(self.fa, self.fa2, t=5, a=3, bwa_path=bwa)
mem.run('output')
with open('output') as fh:
result_output = fh.read().strip()
assert '-t 5' in result_output
assert '-a 3' in result_output
assert self.fa + ' ' + self.fa2 in result_output
def test_requiredargs_gt2(self):
''' Greater than 3 args is not correct '''
mem = BWAMem(self.fa,
self.fa2,
self.fa2,
'arg4',
bwa_path=self.bwa_path)
def bwa_mem( read1, mate=None, ref=None, output='bwa.sai', **kwargs ):
'''
Runs the bwa mem algorithm on read1 against ref. If mate is given then run that file with the read1 file
so paired alignment is done.
TODO:
bwa_path should be an option to specify where the executable is
@param read1 - File path to read
@param mate - Mate file path
@param ref - Reference file path or directory of references
@param output - The output destination
@returns the output path if sucessful or -1 if something went wrong
'''
if os.path.isdir( ref ):
# Compile ref directory
logger.debug( "Compiling references inside of {0}".format(ref) )
ref = compile_refs( ref )
logger.info( "Refs are all compiled into {0}".format(ref) )
# First, make sure the reference is indexed
logger.debug( "Ensuring {0} is indexed".format(ref) )
if not index_ref(ref):
raise InvalidReference("{0} cannot be indexed by bwa")
# Setup BWA Mem
mem = None
if mate:
mem = BWAMem( ref, read1, mate, bwa_path=which_bwa(), **kwargs )
else:
mem = BWAMem( ref, read1, bwa_path=which_bwa(), **kwargs )
ret = mem.run( output )
print "Ret: " + str(ret)
if ret != 0:
return ret
else:
return output
def bwa_mem(read1, mate=None, ref=None, output='bwa.sai', **kwargs):
'''
Runs the bwa mem algorithm on read1 against ref. If mate is given then run that file with the read1 file
so paired alignment is done.
TODO:
bwa_path should be an option to specify where the executable is
@param read1 - File path to read
@param mate - Mate file path
@param ref - Reference file path or directory of references
@param output - The output destination
@returns the output path if sucessful or -1 if something went wrong
'''
if os.path.isdir(ref):
# Compile ref directory
logger.debug("Compiling references inside of {0}".format(ref))
ref = compile_refs(ref)
logger.info("Refs are all compiled into {0}".format(ref))
# First, make sure the reference is indexed
logger.debug("Ensuring {0} is indexed".format(ref))
if not index_ref(ref):
raise InvalidReference("{0} cannot be indexed by bwa")
# Setup BWA Mem
mem = None
if mate:
mem = BWAMem(ref, read1, mate, bwa_path=which_bwa(), **kwargs)
else:
mem = BWAMem(ref, read1, bwa_path=which_bwa(), **kwargs)
ret = mem.run(output)
print "Ret: " + str(ret)
if ret != 0:
return ret
else:
return output
def test_bwareturncode_count(self):
''' Fixed tests for pattern matching '''
# First item should be the sum of read counts
tests = [
(1, '[M::main_mem] read 1 sequences (111350 bp)...'),
(1000,
'[M::main_mem] read 100 sequences (111350 bp)...\n[M::main_mem] read 900 sequences (111350 bp)...'
),
]
for reads, testline in tests:
filename = 'fasta{0}.fa'.format(reads)
util.create_fakefasta(filename, reads)
print "Reads In File: {0}".format(reads)
print "Read Lines: {0}".format(testline)
bwa = BWAMem(self.fa, filename, bwa_path=BWA_PATH)
eq_(0, bwa.bwa_return_code(testline))
def test_bwareturncode_noendline(self):
bwa = BWAMem(self.fa, self.fa2, bwa_path=BWA_PATH)
eq_(1, bwa.bwa_return_code('[main] Version: 0.7.4-r385'))
def test_bwareturncode_nocount(self):
''' Make sure if no read sequences lines exist error is returned '''
endline = '[main] Version: {0}.{1}.{2}-r{3}'
bwa = BWAMem(self.fa, self.fa2, bwa_path=BWA_PATH)
eq_(1, bwa.bwa_return_code(endline.format(0, 7, 4, 385)))
def test_run_nonfastainput(self):
''' Invalid fasta input file(file exists but not fasta/fastq '''
filename = 'test.fa'
with open(filename, 'w') as fh:
fh.write('not a fasta')
bwa = BWAMem(REF_PATH, filename, bwa_path=BWA_PATH, command='mem')
def test_bwamem_run(self):
''' Make sure it actually runs bwa with correct input '''
#infa = ungzip( INPUT_PATH )
bwa.index_ref(self.fa2)
mem = BWAMem(self.fa2, self.fa, bwa_path=BWA_PATH)
eq_(0, mem.run())
def test_requiredargs_third_invalid(self):
''' Test reads file is valid path to file '''
BWAMem(self.fa,
self.fa2,
'/invalid/path/in.fa',
bwa_path=self.bwa_path)
def test_requiredargs_firstarg_nonindexed(self):
''' Test non indexed file for index arg '''
# Should just check to make sure self.fa2 has same filename but with .bwt
shutil.copy(self.fa2, 'noindex.fa')
BWAMem('noindex.fa', self.fa, bwa_path=self.bwa_path)
def test_requiredargs_firstarg_invalid(self):
''' Test invalid path for database arg '''
BWAMem('/invalid/path/in.fa', self.fa2, bwa_path=self.bwa_path)
def test_requiredargs_lt2(self):
''' Needs 2 args '''
mem = BWAMem(self.fa, bwa_path=self.bwa_path)
