def test_find_homologs(self):
"""find_homologs should return raw data, filtered and removed ids."""
formatdb_cmd = 'formatdb -p F -o T -i %s' % self.subjectdb_fp
system(formatdb_cmd)
self._paths_to_clean_up.append("formatdb.log")
for suffix in ["nhr","nin","nsd","nsi","nsq"]:
self._paths_to_clean_up.append(".".join(\
[self.subjectdb_fp,suffix]))
blast_output,hit_ids, removed_hit_ids=\
find_homologs(self.query_fp, self.subjectdb_fp, e_value = 1e-4,
max_hits = 100, working_dir = "./", blast_mat_root = None,\
wordsize = 28, percent_aligned = 0.98, DEBUG=False)
self.assertEqual(hit_ids,set(["bth:BT_0001","hsa:8355"]))
self.assertEqual(removed_hit_ids,set([]))
i=0
for line in blast_output:
if line.startswith("#"):
i+=1
continue #depends on tmpfilename, skip testing
self.assertEqual(blast_output[i],EXP_BLAST_OUTPUT[i])
i+=1
# Ensure low % alignment seqs are removed
blast_output,hit_ids, removed_hit_ids=\
find_homologs(self.query2_fp, self.subjectdb_fp,\
e_value = 1e-4, max_hits = 100, working_dir = "./",\
blast_mat_root = None, wordsize = 28, percent_aligned = 1.00,\
DEBUG=False)
self.assertEqual(hit_ids,set(["bth:BT_0001"]))
self.assertEqual(removed_hit_ids,set(["hsa:8355_tweaked"]))
#Ensure high % alignment seqs are not removed
blast_output,hit_ids, removed_hit_ids=\
find_homologs(self.query2_fp, self.subjectdb_fp, \
e_value = 1e-4, max_hits = 100, working_dir = "./", \
blast_mat_root = None, wordsize = 28, percent_aligned = 0.75,\
DEBUG=False)
self.assertEqual(hit_ids,set(["bth:BT_0001","hsa:8355_tweaked"]))
self.assertEqual(removed_hit_ids,set([]))
def test_find_homologs(self):
"""find_homologs should return raw data, filtered and removed ids."""
formatdb_cmd = 'formatdb -p F -o T -i %s' % self.subjectdb_fp
system(formatdb_cmd)
self._paths_to_clean_up.append("formatdb.log")
for suffix in ["nhr","nin","nsd","nsi","nsq"]:
self._paths_to_clean_up.append(".".join(\
[self.subjectdb_fp,suffix]))
blast_output,hit_ids, removed_hit_ids=\
find_homologs(self.query_fp, self.subjectdb_fp, e_value = 1e-4,
max_hits = 100, working_dir = "./", blast_mat_root = None,\
wordsize = 28, percent_aligned = 0.98, DEBUG=False)
self.assertEqual(hit_ids,set(["bth:BT_0001","hsa:8355"]))
self.assertEqual(removed_hit_ids,set([]))
i=0
for line in blast_output:
if line.startswith("#"):
i+=1
continue #depends on tmpfilename, skip testing
self.assertEqual(blast_output[i],EXP_BLAST_OUTPUT[i])
i+=1
# Ensure low % alignment seqs are removed
blast_output,hit_ids, removed_hit_ids=\
find_homologs(self.query2_fp, self.subjectdb_fp,\
e_value = 1e-4, max_hits = 100, working_dir = "./",\
blast_mat_root = None, wordsize = 28, percent_aligned = 1.00,\
DEBUG=False)
self.assertEqual(hit_ids,set(["bth:BT_0001"]))
self.assertEqual(removed_hit_ids,set(["hsa:8355_tweaked"]))
#Ensure high % alignment seqs are not removed
blast_output,hit_ids, removed_hit_ids=\
find_homologs(self.query2_fp, self.subjectdb_fp, \
e_value = 1e-4, max_hits = 100, working_dir = "./", \
blast_mat_root = None, wordsize = 28, percent_aligned = 0.75,\
DEBUG=False)
self.assertEqual(hit_ids,set(["bth:BT_0001","hsa:8355_tweaked"]))
self.assertEqual(removed_hit_ids,set([]))
def main():
option_parser, options, args = parse_command_line_parameters(**script_info)
DEBUG = options.verbose
check_options(option_parser, options)
start_time = time()
option_lines = format_options_as_lines(options)
if DEBUG:
print FORMAT_BAR
print "Running with options:"
for line in sorted(option_lines):
print line
print FORMAT_BAR
# because the blast app controller uses absolute paths, make sure subject
# db path is fully specified
subject_db = options.subjectdb
if not subject_db.startswith('/'):
subject_db = join(getcwd(), subject_db)
if not options.no_format_db:
# initialize object
inpath = FilePath(abspath(options.subjectdb))
subject_dir, subj_file = split(inpath)
fdb = FormatDb(WorkingDir=subject_dir)
# Currently we do not support protein blasts, but
# this would be easy to add in the future...
fdb.Parameters['-p'].on('F')
# Create indices for record lookup
fdb.Parameters['-o'].on('T')
# Set input database
fdb.Parameters['-i'].on(subject_db)
formatdb_cmd = fdb.BaseCommand
if DEBUG:
print "Formatting db with command: %s" % formatdb_cmd
app_result = fdb(subject_db)
formatdb_filepaths = []
for v in app_result.values():
try:
formatdb_filepaths.append(v.name)
except AttributeError:
# not a file object, so no path to return
pass
db_format_time = time() - start_time
if DEBUG:
print "Formatting subject db took: %2.f seconds" % db_format_time
print "formatdb log file written to: %s" % app_result['log']
print FORMAT_BAR
else:
db_format_time = time() - start_time
formatdb_cmd = "None (formatdb not called)"
# Check that User-Supplied subjectdb is valid
db_ext = [".nhr", ".nin", ".nsd", ".nsi", ".nsq"]
formatdb_filepaths = [subject_db + ext for ext in db_ext]
if DEBUG:
print "Checking that pre-existing formatdb files exist and can be read."
print "Files to be checked:"
for fp in formatdb_filepaths:
print fp
print FORMAT_BAR
try:
formatdb_files = [open(db_f, "U") for db_f in formatdb_filepaths]
[f.close() for f in formatdb_files]
except IOError:
if DEBUG:
print "Cannot open user-supplied database file:", db_f
option_parser.error(
"""Problem with -d and --no_format_db option combination: Cannot open the following user-supplied database file: %s. Consider running without --no_format_db to let formatdb generate these required files""" %
db_f)
if DEBUG:
print "OK: BLAST Database files exist and can be read."
print FORMAT_BAR
# Perform BLAST search
blast_results, hit_ids, removed_hit_ids = find_homologs(options.querydb,
subject_db, options.e_value, options.max_hits,
options.working_dir, options.blastmatroot, options.wordsize,
options.percent_aligned, DEBUG=DEBUG)
blast_time = (time() - start_time) - db_format_time
if DEBUG:
print "BLAST search took: %2.f minute(s)" % (blast_time / 60.0)
print FORMAT_BAR
# Create output folder
outputdir = options.outputdir
try:
makedirs(outputdir)
except OSError:
pass
# Record raw blast results
raw_blast_results_path = join(outputdir, "raw_blast_results.txt")
f = open(raw_blast_results_path, 'w')
f.writelines(blast_results)
f.close()
# Record excluded seqs
excluded_seqs_path = join(outputdir, "matching.fna")
ids_to_seq_file(hit_ids, options.querydb, excluded_seqs_path, "")
# Record included (screened) seqs
included_seqs_path = join(outputdir, "non-matching.fna")
all_ids = ids_from_fasta_lines(open(options.querydb))
included_ids = set(all_ids) - hit_ids
ids_to_seq_file(included_ids, options.querydb, included_seqs_path, "")
log_lines = compose_logfile_lines(start_time, db_format_time, blast_time,
option_lines, formatdb_cmd,
blast_results, options, all_ids,
hit_ids, removed_hit_ids,
included_ids, DEBUG)
log_path = join(outputdir, "sequence_exclusion.log")
if DEBUG:
print "Writing summary to: %s" % log_path
f = open(log_path, 'w')
f.writelines(log_lines)
f.close()
if not options.no_clean:
if DEBUG:
print FORMAT_BAR
print "| Cleanup |"
print FORMAT_BAR
if not options.no_format_db:
if options.verbose:
print "Cleaning up formatdb files:", formatdb_filepaths
remove_files(formatdb_filepaths)
else:
if options.verbose:
print "Formatdb not run...nothing to clean"
def main():
option_parser, options, args = parse_command_line_parameters(**script_info)
DEBUG = options.verbose
check_options(option_parser, options)
start_time = time()
option_lines = format_options_as_lines(options)
if DEBUG:
print FORMAT_BAR
print "Running with options:"
for line in sorted(option_lines):
print line
print FORMAT_BAR
#because the blast app controller uses absolute paths, make sure subject
#db path is fully specified
subject_db = options.subjectdb
if not subject_db.startswith('/'):
subject_db = join(getcwd(), subject_db)
if not options.no_format_db:
#initialize object
inpath = FilePath(abspath(options.subjectdb))
subject_dir, subj_file = split(inpath)
fdb = FormatDb(WorkingDir=subject_dir)
# Currently we do not support protein blasts, but
# this would be easy to add in the future...
fdb.Parameters['-p'].on('F')
# Create indices for record lookup
fdb.Parameters['-o'].on('T')
# Set input database
fdb.Parameters['-i'].on(subject_db)
formatdb_cmd = fdb.BaseCommand
if DEBUG:
print "Formatting db with command: %s" % formatdb_cmd
app_result = fdb(subject_db)
formatdb_filepaths = []
for v in app_result.values():
try:
formatdb_filepaths.append(v.name)
except AttributeError:
# not a file object, so no path to return
pass
db_format_time = time() - start_time
if DEBUG:
print "Formatting subject db took: %2.f seconds" % db_format_time
print "formatdb log file written to: %s" % app_result['log']
print FORMAT_BAR
else:
db_format_time = time() - start_time
formatdb_cmd = "None (formatdb not called)"
# Check that User-Supplied subjectdb is valid
db_ext = [".nhr", ".nin", ".nsd", ".nsi", ".nsq"]
formatdb_filepaths = [subject_db + ext for ext in db_ext]
if DEBUG:
print "Checking that pre-existing formatdb files exist and can be read."
print "Files to be checked:"
for fp in formatdb_filepaths:
print fp
print FORMAT_BAR
try:
formatdb_files = [open(db_f, "U") for db_f in formatdb_filepaths]
[f.close() for f in formatdb_files]
except IOError:
if DEBUG:
print "Cannot open user-supplied database file:", db_f
option_parser.error(
"""Problem with -d and --no_format_db option combination: Cannot open the following user-supplied database file: %s. Consider running without --no_format_db to let formatdb generate these required files"""
% db_f)
if DEBUG:
print "OK: BLAST Database files exist and can be read."
print FORMAT_BAR
# Perform BLAST search
blast_results,hit_ids, removed_hit_ids = find_homologs(options.querydb,\
subject_db, options.e_value,options.max_hits,\
options.working_dir,options.blastmatroot, options.wordsize,\
options.percent_aligned, DEBUG=DEBUG)
blast_time = (time() - start_time) - db_format_time
if DEBUG:
print "BLAST search took: %2.f minute(s)" % (blast_time / 60.0)
print FORMAT_BAR
#Create output folder
outputdir = options.outputdir
try:
makedirs(outputdir)
except OSError:
pass
#Record raw blast results
raw_blast_results_path = join(outputdir, "raw_blast_results.txt")
f = open(raw_blast_results_path, 'w')
f.writelines(blast_results)
f.close()
#Record excluded seqs
excluded_seqs_path = join(outputdir, "matching.fna")
ids_to_seq_file(hit_ids, options.querydb, excluded_seqs_path, "")
#Record included (screened) seqs
included_seqs_path = join(outputdir, "non-matching.fna")
all_ids = ids_from_fasta_lines(open(options.querydb))
included_ids = set(all_ids) - hit_ids
ids_to_seq_file(included_ids, options.querydb, included_seqs_path, "")
log_lines = compose_logfile_lines(start_time, db_format_time, blast_time,\
option_lines,formatdb_cmd,\
blast_results,options,all_ids,\
hit_ids,removed_hit_ids,\
included_ids,DEBUG)
log_path = join(outputdir, "sequence_exclusion.log")
if DEBUG:
print "Writing summary to: %s" % log_path
f = open(log_path, 'w')
f.writelines(log_lines)
f.close()
if not options.no_clean:
if DEBUG:
print FORMAT_BAR
print "| Cleanup |"
print FORMAT_BAR
if not options.no_format_db:
if options.verbose:
print "Cleaning up formatdb files:", formatdb_filepaths
remove_files(formatdb_filepaths)
else:
if options.verbose:
print "Formatdb not run...nothing to clean"
