def qiime_blast_seqs(seqs,
blast_constructor=Blastall,
blast_program='blastn',
blast_db=None,
refseqs=None,
refseqs_fp=None,
blast_mat_root=None,
params={},
WorkingDir=None,
seqs_per_blast_run=1000,
HALT_EXEC=False):
"""Blast list of sequences.
seqs: a list (or object with list-like interace) of (seq_id, seq)
tuples (e.g., the output of MinimalFastaParser)
"""
assert blast_db or refseqs_fp or refseqs, \
'Must provide either a blast_db or a fasta '+\
'filepath containing sequences to build one.'
if refseqs_fp:
blast_db, db_files_to_remove =\
build_blast_db_from_fasta_path(refseqs_fp,output_dir=WorkingDir)
elif refseqs:
blast_db, db_files_to_remove =\
build_blast_db_from_fasta_file(refseqs,output_dir=WorkingDir)
else:
db_files_to_remove = []
params["-d"] = blast_db
params["-p"] = blast_program
blast_app = blast_constructor(params=params,
blast_mat_root=blast_mat_root,
InputHandler='_input_as_seq_id_seq_pairs',
WorkingDir=WorkingDir,
SuppressStderr=True,
HALT_EXEC=HALT_EXEC)
current_seqs = []
blast_results = BlastResult([])
for seq in seqs:
current_seqs.append(seq)
if len(current_seqs) % seqs_per_blast_run == 0:
if blast_results:
blast_results.update(\
BlastResult(blast_app(current_seqs)['StdOut']))
else:
blast_results = BlastResult(blast_app(current_seqs)['StdOut'])
current_seqs = []
# clean-up run: blast the remaining sequences
blast_results.update(\
BlastResult(blast_app(current_seqs)['StdOut']))
remove_files(db_files_to_remove)
return blast_results
def blastn(seqs,
blast_db="nt",
e_value="1e-20",
max_hits=200,
working_dir="/tmp",
blast_mat_root=None,
extra_params={}):
"""
Returns BlastResult from input seqs, using blastn.
Need to add doc string
"""
# set up params to use with blastp
params = {
# matrix
"-M": "BLOSUM62",
# max procs
"-a": "1",
# expectation
"-e": e_value,
# max seqs to show
"-b": max_hits,
# max one line descriptions
"-v": max_hits,
# program
"-p": "blastn"
}
params.update(extra_params)
# blast
blast_res = blast_seqs(seqs,
Blastall,
blast_mat_root=blast_mat_root,
blast_db=blast_db,
params=params,
add_seq_names=False,
WorkingDir=working_dir)
# get prot id map
if blast_res['StdOut']:
lines = [x for x in blast_res['StdOut']]
return BlastResult(lines)
return None
def setUp(self):
self.blast_lines = BLAST_LINES
self.blast_result=BlastResult(self.blast_lines)
self.subjectdb_fp = get_tmp_filename_as_str(\
prefix='ExcludeByBlastTests_',suffix='.fasta')
self.query_fp = get_tmp_filename_as_str(\
prefix='ExcludeByBlastTests_',suffix='.fasta')
self.query2_fp = get_tmp_filename_as_str(\
prefix='ExcludeByBlastTests_',suffix='.fasta')
open(self.subjectdb_fp,"w").writelines(TEST_BLAST_DB_LINES)
open(self.query_fp,"w").writelines(TEST_BLAST_DB_LINES)
open(self.query2_fp,"w").writelines(TEST_BLAST_DB2_LINES)
self._paths_to_clean_up = [self.subjectdb_fp,self.query_fp,\
self.query2_fp]
def setUp(self):
self.blast_lines = BLAST_LINES
self.blast_result = BlastResult(self.blast_lines)
_, self.subjectdb_fp = mkstemp(prefix='ExcludeByBlastTests_',
suffix='.fasta')
close(_)
_, self.query_fp = mkstemp(prefix='ExcludeByBlastTests_',
suffix='.fasta')
close(_)
_, self.query2_fp = mkstemp(prefix='ExcludeByBlastTests_',
suffix='.fasta')
close(_)
open(self.subjectdb_fp, "w").writelines(TEST_BLAST_DB_LINES)
open(self.query_fp, "w").writelines(TEST_BLAST_DB_LINES)
open(self.query2_fp, "w").writelines(TEST_BLAST_DB2_LINES)
self._paths_to_clean_up = [self.subjectdb_fp, self.query_fp,
self.query2_fp]