def _main(args):
if len(args) < 1:
print >> sys.stderr, "usage: <mtx_1> [<mtx2>...] < fasta"
sys.exit(1)
seqname = ""
seq = ""
for x in sys.stdin:
head_re = re.search(">(\S+)", x)
if head_re and not seq:
seqname = head_re.group(1)
continue
elif head_re:
for x in args:
annot = patser_tools.makePatserAnnotation(sequence=seq,
matrix=x)
for x in annot.getAllFeatures():
if x.tags['strand'] == '-':
print "%s\t%s\t%d\t%s\t%f" % (
seqname, x.tags['motif_name'], x.end,
x.tags['strand'], x.tags['score'])
elif x.tags['strand'] == '+':
print "%s\t%s\t%d\t%s\t%f" % (
seqname, x.tags['motif_name'], x.start,
x.tags['strand'], x.tags['score'])
seqname = head_re.group(1)
seq = ""
else:
seq += x.rstrip()
def patSearch(self):
annot = None
try:
annot = patser_tools.makePatserAnnotation(sequence=self.seq,
matrix=self.matrix)
except Exception:
print "warning: Exception for seq %s" % (self.seq)
annot = None
self.annotation = annot
def patSearch(self):
annot = None
#try:
annot = patser_tools.makePatserAnnotation(sequence=self.seq,
matrix=self.matrix)
#except Exception, e:
# print "warning: %s for seq %s" % (type(e),self.seq_name)
# annot = None
self.annotation = annot
def scan_seqs(seqs,matrices):
hits = {}
name = ""
seq = ""
header = ""
results = ""
for s in seqs:
nameres = re.search(">(\S+)",s)
if nameres and not (name == ""):
hits[name] = {'seq' : seq}
name = nameres.group(0)
seq = ""
elif nameres and (name == ""):
#print "1"
name = nameres.group(0)
seq = ""
else:
seq += s[:-1]
#print hits
mtx_names = []
for (name,d) in hits.iteritems():
for mtx in matrices:
hit_annot = patser_tools.makePatserAnnotation(sequence=d['seq'],matrix=mtx,seqname=name,scorecut=-100)
features = hit_annot.getAllFeatures()
hit = None
if len(features) > 0:
max = features[0]
for x in features:
if x.tags['score'] > max.tags['score']:
max = x
hit = max
else:
print >> sys.stderr, "No hit for matrix %s in %s" % (mtx,d['seq'])
continue
#print hit
d[hit.tags['motif_name']] = hit
if hit.tags['motif_name'] not in mtx_names:
mtx_names.append(hit.tags['motif_name'])
#print hits
header = "name\tsequence\t",
for x in mtx_names:
header += "%s_score\t%s_pval\t" % (x,x),
for (name,h) in hits.iteritems():
matrices = [x for x in h.keys() if not x == 'seq']
result += "%s\t%s\t" % (name,h['seq']) ,
for x in matrices:
result += str(h[x].tags['score']) + "\t",
if 'pval' in h[x].tags.keys():
result += str(h[x].tags['pval']) + "\t",
else:
result += "-",
return (header,result)
def _main(args):
if len(args) < 4:
print >> sys.stderr, "usage: xls_motif_window.py <xls> <fasta> <matrix_file> <window>"
sys.exit(1)
fasta = fasta_subseq_2.FastaDB()
fasta.openFastaFile(args[1])
xls_regions = []
for x in open(args[0]):
spl = x[:-1].split()
region = {
'chr': spl[0],
'start': int(spl[1]),
'end': int(spl[2]),
'enrich': spl[7]
}
region['seq'] = fasta[
region['chr']]['sequence'][region['start']:region['end']]
xls_regions.append(region)
for r in xls_regions:
try:
annot = patser_tools.makePatserAnnotation(sequence=r['seq'],
matrix=args[2])
except IOError:
print >> sys.stderr, "Error in seq %s:%d..%d:" % (
r['chr'], r['start'], r['end'])
continue
if len(annot.getAllFeatures()) < 1:
continue
maxhit = annot.getMaxFeature("score")
winstart = None
winend = None
winseq = None
if maxhit.tags["strand"] == '+':
winstart = r['start'] + (maxhit.start - int(args[3]) / 2)
winend = r['start'] + (maxhit.start + int(args[3]) / 2)
win_seq = fasta[r['chr']]['sequence'][winstart:winend]
else:
winstart = r['start'] + ((maxhit.end - 3) - int(args[3]) / 2)
winend = r['start'] + ((maxhit.end - 3) + int(args[3]) / 2)
win_seq = fasta_subseq_2.revcomp(
fasta[r['chr']]['sequence'][winstart:winend])
print ">%s:%d..%d:%s enr=%s mtx=%s" % (
r['chr'], winstart, winend, maxhit.tags['strand'], r['enrich'],
maxhit.tags['score'])
print win_seq
def _main(args):
if len(args) != 3:
print "usage: <bed_file> <seq_file> <matrix>"
sys.exit(0)
fasta = fasta_subseq_2.FastaDB()
fasta.openFastaFile(args[1])
bed_annots = []
bed_in = open(args[0])
for line in bed_in:
spl = line[:-1].split()
fseq = fasta[spl[0]]["sequence"][int(spl[1]):int(spl[2])]
if spl[5] == "-":
fseq = fasta_subseq_2.revcomp(fseq)
#print spl
try:
patannot = patser_tools.makePatserAnnotation(sequence=fseq,
matrix=args[2])
except:
continue
#print "-" * 30
#print spl
#print pp(patannot.getAllFeatures())
bed_annots.append({
"seq": spl[0] + "_" + spl[1] + "_" + spl[2],
"annotation": patannot
})
for ann in bed_annots:
for feat in ann["annotation"].getAllFeatures():
print "%s\t%i\t%i\t%f\t%f\t%s" % (
ann["seq"], feat.st, feat.en, feat.tagset["score"],
feat.tagset["pval"], feat.tagset["strand"])
def _main(args):
if len(args) < 1:
print "usage: patser_list.py [mtx1][mtx2] ... < seqs.fa"
sys.exit(1)
matrices = args
seqs = sys.stdin
hits = {}
name = ""
seq = ""
pssms = convertFreqMtx(matrices)
for s in seqs:
nameres = re.search(">(\S+)", s)
if nameres and not (name == ""):
hits[name] = {'seq': seq}
name = nameres.group(1)
seq = ""
elif nameres and (name == ""):
#print "1"
name = nameres.group(1)
seq = ""
else:
seq += s[:-1]
if not (name == ""):
hits[name] = {'seq': seq}
#print hits
mtx_names = []
for (name, d) in hits.iteritems():
for mtx in matrices:
hit_annot = patser_tools.makePatserAnnotation(sequence=d['seq'],
matrix=mtx,
seqname=name,
scorecut=-100)
features = hit_annot.getAllFeatures()
hit = None
if len(features) > 0:
max = features[0]
for x in features:
if x.tags['score'] > max.tags['score']:
max = x
hit = max
else:
print >> sys.stderr, "Sequence %s: No hit for matrix %s in %s" % (
name, mtx, d['seq'])
continue
#print hit
d[hit.tags['motif_name']] = hit
if hit.tags['motif_name'] not in mtx_names:
mtx_names.append(hit.tags['motif_name'])
#print hits
print "name\tsequence\t",
for x in mtx_names:
print "%s_score\t%s_pval\t%s_PSSM_score" % (x, x, x),
print ""
for (name, h) in hits.iteritems():
matrices = [x for x in h.keys() if not x == 'seq']
print "%s\t%s\t" % (name, h['seq']),
for x in matrices:
print str(h[x].tags['score']) + "\t",
if 'pval' in h[x].tags.keys():
print str(h[x].tags['pval']) + "\t",
else:
print "-",
pssm_scores = [
scoreSeq(pssms[x], h['seq']),
scoreSeq(pssms[x], fasta_subseq_2.revcomp(h['seq']))
]
score = None
if pssm_scores[1] > pssm_scores[0]:
score = pssm_scores[1]
else:
score = pssm_scores[0]
print str(score) + "\t",
print ""