def MOODS_search(seq, motif, thresholds=0):
"""an equivalent to Motif.search_pwm()"""
if not USE_MOODS:
raise RuntimeError("MOODS could not be imported")
sequence = seq
matrix_ = MOODS.transpose([map(lambda x: x[1], sorted(x.items())) for x in motif.log_odds()])
# Note: algorithm = 'lf' fails due to segmentation fault
results_per_matrix = MOODS.search(
sequence,
[matrix_],
thresholds,
bg=None,
algorithm="pla",
q=7,
absolute_threshold=True,
both_strands=True,
combine=True,
)
# format as Motif.search_pwm results
search_results = results_per_matrix[0]
# figure out direction of reverse results
# do we need to reverse it?
results_sorted_like_Bio_Motif = sorted(
search_results, key=operator.itemgetter(0), cmp=lambda x, y: cmp(abs(x), abs(y))
)
return results_sorted_like_Bio_Motif
def getMOODSscore_old(seq, mat):
results = MOODS.search(seq, [mat], thresholds=1, absolute_threshold=False)
resarray = np.zeros((1, len(seq)), dtype=np.dtype('Float32'))
resarray[:, :] = -100
for (position, score) in results[0]:
resarray[0, position] = score
return (resarray)
def getMOODSscore(seqfile, pwmfiles, both_strands=False):
handle = open(seqfile, "r")
records = list(Bio.SeqIO.parse(handle, "fasta"))
handle.close()
seq = records[0].seq
print 'len(seq)=', len(seq)
matrixlist = list()
for f in pwmfiles:
matrix = MOODS.load_matrix(f)
print 'pwm ', f, 'windowlength=', len(matrix[0])
matrixlist.append(matrix)
if both_strands:
matrixlist.append(
MOODS.reverse_complement(matrix)
) # both_strand option in MOODS returned a akward result.
print 'starting MOODS.search', datetime.now()
results = MOODS.search(seq,
matrixlist,
thresholds=1,
absolute_threshold=False)
print 'done MOODS.search', datetime.now()
reslist = []
for n in range(len(pwmfiles)):
thisind = n * (1 + both_strands)
reslist.append(vegardparseMOODSres(results[thisind], len(seq)))
if both_strands:
reslist[n] = np.append(reslist[n],
vegardparseMOODSres(results[thisind + 1],
len(seq)),
axis=0)
return (reslist)
def getPWMscores(regions, PWM, fasta, regionsFile):
"""
Score every basepairof a set of regions with a match to a PWM.
:param regions: `bedtools.BedTool` with regions to look at.
:type regions: bedtools.BedTool
:param PWM:
:type PWM:
:param fasta: Fasta file with genome sequence.
:type fasta: str
:param regionsFile:
:type regionsFile:
"""
import MOODS
import numpy as np
import pandas as pd
# Get nucleotides
seq = regions.sequence(s=True, fi=fasta)
with open(seq.seqfn) as handle:
seqs = handle.read().split("\n")[1::2] # get fasta sequences
# Match strings with PWM
scores = list()
for sequence in seqs:
result = MOODS.search(sequence, [PWM], 30)
scores.append(np.array([j for i, j in result[0]]))
names = open(regionsFile).read().split("\n")
names.pop(-1)
names = [line.split("\t")[3] for line in names]
return pd.DataFrame(scores, index=names, columns=range(-990, 990))
def getMOODSscore_old(seq, mat):
results = MOODS.search(seq, [mat], thresholds=1, absolute_threshold=False)
resarray = np.zeros( (1, len(seq)), dtype=np.dtype('Float32'))
resarray[:,:] = -100
for (position, score) in results[0]:
resarray[0, position] = score
return(resarray)
def getPWMscores(regions, PWM, fasta, regionsFile):
"""
Score every basepairof a set of regions with a match to a PWM.
:param regions: `bedtools.BedTool` with regions to look at.
:type regions: bedtools.BedTool
:param PWM:
:type PWM:
:param fasta: Fasta file with genome sequence.
:type fasta: str
:param regionsFile:
:type regionsFile:
"""
import MOODS
import numpy as np
import pandas as pd
# Get nucleotides
seq = regions.sequence(s=True, fi=fasta)
with open(seq.seqfn) as handle:
seqs = handle.read().split("\n")[1::2] # get fasta sequences
# Match strings with PWM
scores = list()
for sequence in seqs:
result = MOODS.search(sequence, [PWM], 30)
scores.append(np.array([j for i, j in result[0]]))
names = open(regionsFile).read().split("\n")
names.pop(-1)
names = [line.split("\t")[3] for line in names]
return pd.DataFrame(scores, index=names, columns=range(-990, 990))
def search(consensus_list,TF,master,search_region,p): count = 0 count1 = 0 interaction_only = False try: if sys.argv[3] == "True": interaction_only = True else: interaction_only = False except IndexError: pass duplicate = [] threshold = [] header = '' for i in master: count +=1 threshold += [p] print >> sys.stderr, count #print threshold for region in search_region: if 'strand' in region: continue if '>' in region: header = region else: for i in range(len(master)): tf = TF[i] #print tf,consensus_list[i] tf_length = len(consensus_list[i]) #result = MOODS.search(region,master,threshold,absolute_threshold=threshold) result = MOODS.search(region,master,p) for check in range(len(result)): for j in range(len(result[check])): position = result[check][j][0] tf_length = len(consensus_list[check]) if result[check][j] == []: continue if interaction_only: if [TF[check],header] not in duplicate: duplicate += [[TF[check],header]] print tf.strip(),header.strip()[1:] print '' else: if [TF[check],header,position] not in duplicate: duplicate+= [[TF[check],header,position]] print TF[check].strip(),header.strip()[1:] print 'position:',position align(consensus_list[check],region[position:position+tf_length]) #print consensus_list[i],'Matched the motif in the upstream region:',region[position:position+tf_length] print ''
def ProcessSeqs(SEQ_HANDLE, PWMS, THRESHOLD, WANT_REV=False, bg=None):
"""Yields matches on sequences in an 'interval' formatted dictionary"""
pwm_names = map(lambda x: x[0], PWMS)
pwm_mats = map(lambda x: x[1], PWMS)
thresh = map(lambda x: MOODS.threshold_from_p(x, bg, THRESHOLD), pwm_mats)
for interval in ReadInterval(SEQ_HANDLE):
print interval['NAME']
results = MOODS.search(interval['SEQ'].upper(),
pwm_mats,
thresh,
both_strands=WANT_REV,
algorithm='lf',
absolute_threshold=True,
bg=bg)
for res, pwm_name, pwm_mat, th in zip(results, pwm_names, pwm_mats,
thresh):
width = len(pwm_mat[0])
for position, score in res:
if score > th:
yield {
'NAME': interval['NAME'],
'START': int(interval['START']) + position,
'END': int(interval['START']) + width + position,
'STRAND': interval['STRAND'],
'PWM': pwm_name,
'SCORE': score,
'CHROM': interval['CHROM'],
'SEQ':
interval['SEQ'][position:(position + width)].upper()
}
else:
print 'got bad result'
def getMOODSscore(seqfile, pwmfiles, both_strands=False):
handle = open(seqfile, "r")
records = list(Bio.SeqIO.parse(handle, "fasta"))
handle.close()
seq = records[0].seq
print 'len(seq)=',len(seq)
matrixlist=list()
for f in pwmfiles:
matrix = MOODS.load_matrix(f)
print 'pwm ', f , 'windowlength=', len(matrix[0])
matrixlist.append(matrix)
if both_strands:
matrixlist.append(MOODS.reverse_complement(matrix)) # both_strand option in MOODS returned a akward result.
print 'starting MOODS.search', datetime.now()
results = MOODS.search(seq, matrixlist, thresholds=1, absolute_threshold=False)
print 'done MOODS.search', datetime.now()
reslist=[]
for n in range(len(pwmfiles)):
thisind = n * (1 + both_strands)
reslist.append(vegardparseMOODSres( results[thisind] , len(seq)))
if both_strands:
reslist[n] = np.append( reslist[n] , vegardparseMOODSres( results[thisind+1] , len(seq)), axis=0)
return(reslist)
records = fasta.parseFasta(fasta_filepath)
seq = records[0][1]
matrix1 = [ [0,1,0,0,0,0,0,1,1,0],
[1,0,0,0,0,0,0,0,0,0],
[0,0,0,0,0,0,0,0,0,0],
[0,0,1,1,1,1,1,0,0,1]
]
matrix2 = [ [10,0,10,3,5,5],
[0,5,0,3,5,0,5],
[0,1,0,3,0,5,0],
[0,4,0,1,0,0,5]
]
results = MOODS.search(seq, [matrix1, matrix2], 0.011)
print("Matrix 1 results: "+ str(len(results[0])))
print("Matrix 2 results: "+ str(len(results[1])))
matrices = [matrix1, matrix2]
thresholds = [0.011, 0.011]
bg = MOODS.bg_from_sequence(seq, 0.1)
q = 7
absolute_threshold = False
both_strands=False
ms = MOODS.MOODSSearch(matrices, thresholds, bg, q, absolute_threshold, both_strands)
results = ms.search(seq)
print("New Matrix 1 results: "+ str(len(results[0])))
def main():
"""
The main loop. Lets ROCK!
"""
desc = """... ask me later! I'm on a deadline! ..."""
parser = argparse.ArgumentParser(description=desc)
parser.add_argument('--seqs', type=str,
help="""Path to a fasta file containing the 'promoter'
regions of a single species you wish to scan with motifs.""")
parser.add_argument('--species', type=str,
help="""A quoted string of the species name: 'Anophles gambiae'.""")
parser.add_argument('--motifs', type=str,
help="""Path to a file containing the motifs you wish to use.
The file must be in JASPAR's 'matrix_only.txt' format.""")
parser.add_argument('--thresh', type=float, required=False, default=0.001,
help="""A p-val cut-off above which hits will be ignored. (default = %(default)s)""")
parser.add_argument('--out', type=str, required=False, default='compare_motifs.out',
help="""Path to outfile. (default = %(default)s)""")
parser.add_argument('--norm', type=str, required=False, default=False,
help="""Optional path to outfile for data normalized by upper quartiles w.r.t. each motif. (default = %(default)s)""")
parser.add_argument('--to-norm', type=str, required=False, default=False,
help="""Optional path to outfile of previous run that needs to be normalized. (default = %(default)s)""")
args = parser.parse_args()
if not args.to_norm:
# create parsers
motifs = ParseJasparMatrixOnly(args.motifs)
seqs = ParseFastA(args.seqs)
# Load all motifs at once
# We will be loading one seq at a time.
motifs = motifs.to_dict()
# set up output and headers
headers = 'seq_name\tspecies\t%s\n' % ('\t'.join(motifs.keys()))
out_file = open(args.out,'w')
out_file.write(headers)
# lets start the major looping
if args.norm and not args.to_norm:
norm_dict = OrderedDict()
elif args.to_norm:
norm_dict,headers = load_to_normalize(args.to_norm)
write_normalized_table(headers,args.norm,norm_dict)
exit(0)
for name,seq in seqs:
hits = MOODS.search(seq,motifs.values(),args.thresh)
counts = [len(x) for x in hits]
out_file.write('%s\t%s\t%s\n' % (name,args.species,'\t'.join([str(x) for x in counts])))
if args.norm:
norm_dict['%s\t%s' % (name,args.species)] = np.array(counts)
out_file.close()
if args.norm:
write_normalized_table(headers,args.norm,norm_dict)
[0,0,0,1]]
teststring = 'acgtacgt'
'''
handle = open('/xanadu/home/vegardny/prosjekter/hyperbrowser/pwm_vs_snp/vegard_debug_MOODS/examples/vegardtest2.fasta', "r")
records = list(Bio.SeqIO.parse(handle, "fasta"))
handle.close()
seq = records[0]
teststring=seq.seq
print('both strands')
results = MOODS.search(teststring, [matrix], thresholds=1, absolute_threshold=False, both_strands=True)
for i in results:
for (position, score) in i:
print("Position: " + str(position) + " Score: "+ str(score))
print('one way')
results = MOODS.search(teststring, [matrix], thresholds=1, absolute_threshold=False, both_strands=False)
for i in results:
for (position, score) in i:
print("Position: " + str(position) + " Score: "+ str(score))
import MOODS
matrix = [ [10,0,0],
[0,10,0],
[0,0,10],
[10,10,10]]
results = MOODS.search('actgtggcgtcaacgtaggccaacgtggacccgtacgtaaacgaagaggggtagtc', [matrix], 30, absolute_threshold=30)
for i in results:
for (position, score) in i:
print("Position: " + str(position) + " Score: "+ str(score))
def find_motif_disruptions(
position,
ref,
alt,
genome_fasta,
matrices,
):
"""
Determine whether there is a difference between the ref and alt
alleles for TF binding. Requires samtools in your path.
Parameters
----------
position : str
Zero based genomic coordinates of the reference allele of the form
chrom:start-end (chr5:100-101 for a SNV for instance). The value end -
start should equal the length of the ref allele.
ref : str
Reference allele. This should match the reference sequence at "position"
in genome_fasta.
alt : str
Alternate allele.
genome_fasta : str
Path to genome fasta file. This file should be indexed.
matrices : dict
Dict whose keys are motif names and whose values are pandas data frames
or numpy arrays containing PWMs with columns ACGT.
Returns
-------
out : pandas.DataFrame
Pandas data frame with motifs whose best matches that overlapped the
variant differed between the reference and alternate sequences. A score
of zero and a strand of '' indicates that there was not a match for the
motif on the given allele.
"""
import subprocess
import MOODS
# import pybedtools as pbt
max_motif_length = max([x.shape[0] for x in matrices.values()])
chrom, coords = position.split(':')
start, end = [int(x) for x in coords.split('-')]
s = '{}:{}-{}'.format(chrom, start - max_motif_length + 1,
end + max_motif_length - 1)
c = 'samtools faidx {} {}'.format(genome_fasta, s)
seq_lines = subprocess.check_output(c, shell=True).strip().split()
ref_seq = seq_lines[1]
alt_seq = ref_seq[0:max_motif_length -
1] + alt + ref_seq[max_motif_length + len(ref) - 1:]
ref_variant_start = max_motif_length - 1
ref_variant_end = max_motif_length - 1 + len(ref)
alt_variant_start = max_motif_length - 1
alt_variant_end = max_motif_length - 1 + len(alt)
ms = [matrices[x].T.values.tolist() for x in matrices.keys()]
ref_res = MOODS.search(ref_seq,
ms,
0.001,
both_strands=True,
bg=[0.25, 0.25, 0.25, 0.25])
ref_res = dict(zip(matrices.keys(), ref_res))
alt_res = MOODS.search(alt_seq,
ms,
0.001,
both_strands=True,
bg=[0.25, 0.25, 0.25, 0.25])
alt_res = dict(zip(matrices.keys(), alt_res))
# First we'll remove any motif matches that don't overlap the variant of interest (and thus
# can't be affected by the variant and will be the same for ref and alt). Then we'll get the
# best match for each motif for ref and alt.
rows = []
for motif in ref_res.keys():
ref_res[motif] = _filter_variant_motif_res(ref_res[motif],
ref_variant_start,
ref_variant_end,
matrices[motif].shape[0],
ref_seq)
alt_res[motif] = _filter_variant_motif_res(alt_res[motif],
alt_variant_start,
alt_variant_end,
matrices[motif].shape[0],
alt_seq)
if len(ref_res[motif]) > 0:
ref_pos, ref_score = sorted(ref_res[motif],
key=lambda x: x[1],
reverse=True)[0]
ref_strand = {True: '+', False: '-'}[ref_pos > 0]
else:
ref_score = 0
ref_strand = ''
if len(alt_res[motif]) > 0:
alt_pos, alt_score = sorted(alt_res[motif],
key=lambda x: x[1],
reverse=True)[0]
alt_strand = {True: '+', False: '-'}[alt_pos > 0]
else:
alt_score = 0
alt_strand = ''
if ref_score > 0 or alt_score > 0:
diff = ref_score - alt_score
rows.append(
[motif, ref_score, ref_strand, alt_score, alt_strand, diff])
out = pd.DataFrame(rows,
columns=[
'motif', 'ref_score', 'ref_strand', 'alt_score',
'alt_strand', 'score_diff'
])
out.index = out.motif
out = out.drop('motif', axis=1)
out = out[out.score_diff != 0]
return out
def count(enhlist, c, p_val):
write_match = False
index = 0
seq = ""
while index <= len(enhlist) - 1 and not enhlist[index].startswith('>'):
seq += enhlist[index]
seq = seq.replace('\n','')
index += 1
index = 0
results = []
pseudocount = 0.001
# starrmot15
# caca = [[49,0,100,0,91,0,100,0],[1,83,0,88,0,49,0,94],[49,1,0,12,0,9,0,6],[0,16,0,0,9,41,0,0,]]
caca = [[0,1,0,1,0,1],[1,0,1,0,1,0],[0,0,0,0,0,0],[0,0,0,0,0,0]]
me137 = [[0,1,0,1,0,1],[0,0,0,0,0,0],[1,0,1,0,1,0],[0,0,0,0,0,0]]
gcgc = [[0,0,0,0,0,0],[0,1,0,1,0,1],[1,0,1,0,1,0],[0,0,0,0,0,0]]
tata = [[0,1,0,1,0,1],[0,0,0,0,0,0],[0,0,0,0,0,0],[1,0,1,0,1,0]]
matrices = [me137,caca,gcgc,tata]
if (c == "default"):
results = MOODS.search(seq,matrices,p_val,both_strands=True)
elif (c == "human"):
BG_hum =[0.29508855202553025, 0.20466109233964447, 0.20478482916547036, 0.2954655264693549]
matrices = [MOODS.count_log_odds(matrix,BG_hum,pseudocount) for matrix in matrices]
thresholds = [MOODS.threshold_from_p(matrix, BG_hum, p_val) for matrix in matrices]
results = MOODS.search(seq, matrices, thresholds, convert_log_odds=False, threshold_from_p=False,both_strands=True)
elif (c == "equal"):
BG_eq = [0.25,0.25,0.25,0.25]
matrices = [MOODS.count_log_odds(matrix,BG_eq,pseudocount) for matrix in matrices]
thresholds = [MOODS.threshold_from_p(matrix, BG_eq, p_val) for matrix in matrices]
results = MOODS.search(seq, matrices, thresholds, convert_log_odds=False, threshold_from_p=False,both_strands=True)
else:
write_match = False
ga = re.compile(r'GAGA(?:GA)+',re.IGNORECASE)
ct = re.compile(r'TCTC(?:TC)+',re.IGNORECASE)
ca = re.compile(r'CACA(?:CA)+',re.IGNORECASE)
gt = re.compile(r'TGTG(?:TG)+',re.IGNORECASE)
gc = re.compile(r'GCGC(?:GC)+',re.IGNORECASE)
cg = re.compile(r'CGCG(?:CG)+',re.IGNORECASE)
ta = re.compile(r'TATA(?:TA)+',re.IGNORECASE)
at = re.compile(r'ATAT(?:AT)+',re.IGNORECASE)
ga_occur = re.findall(ga,seq) + re.findall(ct,seq)
ca_occur = re.findall(ca,seq) + re.findall(gt,seq)
gc_occur = re.findall(gc,seq) + re.findall(cg,seq)
ta_occur = re.findall(ta,seq) + re.findall(at,seq)
results = [ga_occur, ca_occur, gc_occur, ta_occur]
# write files containing the matches for each motif as found by MOODS
# only if boolean set at top of count() is True
if write_match:
ga = open('%sga_matches_%d.txt' % (time.strftime("%Y-%m-%d"),1/p_val), 'a')
ca = open('%sca_matches_%d.txt' % (time.strftime("%Y-%m-%d"),1/p_val), 'a')
gc = open('%sgc_matches_%d.txt' % (time.strftime("%Y-%m-%d"),1/p_val), 'a')
ta = open('%sta_matches_%d.txt' % (time.strftime("%Y-%m-%d"),1/p_val), 'a')
comp = False
for match in results[0]:
pos = match[0]
st = ""
if (pos < 0):
comp = True
pos += len(seq)
for i in range(len(me137[1])):
st += seq[pos+i]
if comp:
st = complement(st)
comp = False
ga.write('%s\n' % st)
for match in results[1]:
pos = match[0]
st = ""
if (pos < 0):
comp = True
pos += len(seq)
for i in range(len(caca[1])):
st += seq[pos+i]
if comp:
st = complement(st)
comp = False
ca.write('%s\n' % st)
for match in results[2]:
pos = match[0]
st = ""
if (pos < 0):
comp = True
pos += len(seq)
for i in range(len(gcgc[1])):
st += seq[pos+i]
if comp:
st = complement(st)
comp = False
gc.write('%s\n' % st)
for match in results[3]:
pos = match[0]
st = ""
if (pos < 0):
comp = True
pos += len(seq)
for i in range(len(tata[1])):
st += seq[pos+i]
if comp:
st = complement(st)
comp = False
ta.write('%s\n' % st)
ga.close()
ca.close()
gc.close()
ta.close()
results.append(len(seq))
return results
def match_single(motif,
sequence,
genomic_region,
unique_threshold=None,
normalize_bitscore=True,
sort=False):
"""
Performs motif matching given sequence and the motif.pssm passed as parameter.
The genomic_region is needed to evaluate the correct binding position.
Please note that the arguments should be passed as a list, to allow for parallelization
mapping function.
Keyword arguments:
motif -- TODO.
sequence -- A DNA sequence (string).
genomic_region -- A GenomicRegion.
output_file -- TODO.
unique_threshold -- If this argument is provided, the motif search will be made using a threshold of 0 and
then accepting only the motif matches with bitscore/motif_length >= unique_threshold.
Return:
Print MPBSs to output_file.
"""
# Establishing threshold
if unique_threshold:
current_threshold = 0.0
eval_threshold = unique_threshold
motif_max = motif.max / motif.len
else:
current_threshold = motif.threshold
eval_threshold = motif.threshold
motif_max = motif.max
# Performing motif matching
try:
# old MOODS version
results = MOODS.search(sequence, [motif.pssm_list],
current_threshold,
absolute_threshold=True,
both_strands=True)
except:
# TODO: we can expand this to use bg from sequence, for example,
# or from organism.
bg = MOODS.tools.flat_bg(4)
results = MOODS.scan.scan_dna(sequence, [motif.pssm_list], bg,
[current_threshold], 7)
grs = GenomicRegionSet("mpbs")
for search_result in results:
for r in search_result:
try:
position = r.pos
score = r.score
except:
(position, score) = r
# Verifying unique threshold acceptance
if unique_threshold and score / motif.len < unique_threshold:
continue
# If match forward strand
if position >= 0:
p1 = genomic_region.initial + position
strand = "+"
# If match reverse strand
elif not motif.is_palindrome:
p1 = genomic_region.initial - position
strand = "-"
else:
continue
# Evaluating p2
p2 = p1 + motif.len
# Evaluating score (integer between 0 and 1000 -- needed for bigbed transformation)
if normalize_bitscore:
# Normalized bitscore = standardize to integer between 0 and 1000 (needed for bigbed transformation)
if motif_max > eval_threshold:
norm_score = int(((score - eval_threshold) * 1000.0) /
(motif_max - eval_threshold))
else:
norm_score = 1000
else:
# Keep the original bitscore
if unique_threshold:
norm_score = score / motif.len
else:
norm_score = score
grs.add(
GenomicRegion(genomic_region.chrom,
int(p1),
int(p2),
name=motif.name,
orientation=strand,
data=str(norm_score)))
if sort:
grs.sort()
return grs
import MOODS
matrix = [[10, 0, 0], [0, 10, 0], [0, 0, 10], [10, 10, 10]]
results = MOODS.search(
'actgtggcgtcaacgtaggccaacgtggacccgtacgtaaacgaagaggggtagtc', [matrix],
30,
absolute_threshold=30)
for i in results:
for (position, score) in i:
print("Position: " + str(position) + " Score: " + str(score))
def main():
p = optparse.OptionParser(__doc__)
p.add_option('-t',
'--thresh',
action='store',
dest='threshold',
default=0.0,
help='determines threshold')
p.add_option('-a',
'--append',
action='store',
dest='name',
default='resultsfor',
help='appends pwm name to this when\
creating files')
p.add_option('-A',
'--absolute',
action='store_true',
dest='A',
default=False,
help='absolute threshold')
p.add_option('-s',
'--standard_background',
action='store_true',
dest='stdbg')
p.add_option('-M', '--specific_Matrix', action='store', dest='specific')
options, args = p.parse_args()
pwm = open(args[0], 'rU')
fa = open(args[1], 'rU')
pfa = list(Bio.SeqIO.parse(fa, 'fasta'))
index, matricies, sizes = pySeq.parsing.PWMparser.parse(pwm)
underorequal20 = []
over20 = []
under20names = []
over20names = []
pwmdata = {}
fileout = {}
bgt = False
if options.stdbg:
bgt = [0.25, 0.25, 0.25, 0.25]
# Construct Matrices to search and files to write to.
for k in index.keys():
if options.specific:
if k == options.specific:
filename = options.name + k + '.bed'
fileout[k] = open(filename, 'w')
if sizes[k] <= 20:
underorequal20.append(matricies[k])
under20names.append(k)
else:
over20.append(matricies[k])
over20names.append(k)
else:
filename = options.name + k + '.bed'
fileout[k] = open(filename, 'w')
if sizes[k] <= 20:
underorequal20.append(matricies[k])
under20names.append(k)
else:
over20.append(matricies[k])
over20names.append(k)
for chrom in pfa:
print(chrom.name)
#Run under 20s
# Should we sort the results as all downstream applications require a
# sort first
res = MOODS.search(chrom.seq,
underorequal20,
float(options.threshold),
absolute_threshold=options.A,
both_strands=True,
bg=bgt,
algorithm='lf')
for n, r in enumerate(res):
for position, score in r:
start, end, strand = strand_adjust(position,
sizes[under20names[n]])
# Add option to round the score values. Defaulting to int atm
# since bedToBigBed only accepts integer values....
fileout[under20names[n]].write('\t'.join([
chrom.name,
str(start),
str(end), under20names[n],
str(int(score * 100)), strand, '\n'
]))
#Run over 20s
res = MOODS.search(chrom.seq,
over20,
float(options.threshold),
absolute_threshold=options.A,
both_strands=True,
bg=bgt,
algorithm='supera')
for n, r in enumerate(res):
for position, score in r:
start, end, strand = strand_adjust(position,
sizes[over20names[n]])
fileout[over20names[n]].write('\t'.join([
chrom.name,
str(start),
str(end), over20names[n],
str(int(score * 100)), strand, '\n'
]))
import fasta
DIST_DIR = abspath(dirname(dirname(LOCAL_DIR)))
print(DIST_DIR)
fasta_filepath = join(DIST_DIR, "examples/data/sequence/dnaACGT.txt")
records = fasta.parseFasta(fasta_filepath)
seq = records[0][1]
matrix1 = [[0, 1, 0, 0, 0, 0, 0, 1, 1, 0], [1, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 1, 1, 1, 1, 1, 0, 0, 1]]
matrix2 = [[10, 0, 10, 3, 5, 5], [0, 5, 0, 3, 5, 0, 5], [0, 1, 0, 3, 0, 5, 0],
[0, 4, 0, 1, 0, 0, 5]]
results = MOODS.search(seq, [matrix1, matrix2], 0.011)
print("Matrix 1 results: " + str(len(results[0])))
print("Matrix 2 results: " + str(len(results[1])))
matrices = [matrix1, matrix2]
thresholds = [0.011, 0.011]
bg = MOODS.bg_from_sequence(seq, 0.1)
q = 7
absolute_threshold = False
both_strands = False
ms = MOODS.MOODSSearch(matrices, thresholds, bg, q, absolute_threshold,
both_strands)
results = ms.search(seq)
print("New Matrix 1 results: " + str(len(results[0])))
def main():
p = optparse.OptionParser(__doc__)
p.add_option('-t', '--thresh', action='store', dest='threshold',
default=0.0,help='determines threshold')
p.add_option('-a', '--append', action='store', dest='name',
default='resultsfor', help='appends pwm name to this when\
creating files')
p.add_option('-A', '--absolute', action='store_true',dest='A',
default=False,help='absolute threshold')
p.add_option('-s','--standard_background',action='store_true',dest='stdbg')
p.add_option('-M', '--specific_Matrix', action='store', dest='specific')
options, args = p.parse_args()
pwm = open(args[0], 'rU')
fa = open(args[1], 'rU')
pfa = list(Bio.SeqIO.parse(fa, 'fasta'))
index, matricies, sizes = pySeq.parsing.PWMparser.parse(pwm)
underorequal20 = []
over20 = []
under20names = []
over20names = []
pwmdata={}
fileout = {}
bgt = False
if options.stdbg:
bgt = [0.25,0.25,0.25,0.25]
# Construct Matrices to search and files to write to.
for k in index.keys():
if options.specific:
if k == options.specific:
filename = options.name + k + '.bed'
fileout[k] = open(filename, 'w')
if sizes[k] <= 20:
underorequal20.append(matricies[k])
under20names.append(k)
else:
over20.append(matricies[k])
over20names.append(k)
else:
filename = options.name + k + '.bed'
fileout[k] = open(filename, 'w')
if sizes[k] <= 20:
underorequal20.append(matricies[k])
under20names.append(k)
else:
over20.append(matricies[k])
over20names.append(k)
for chrom in pfa:
print(chrom.name)
#Run under 20s
# Should we sort the results as all downstream applications require a
# sort first
res = MOODS.search(chrom.seq, underorequal20, float(options.threshold),
absolute_threshold=options.A , both_strands = True,
bg=bgt, algorithm='lf')
for n,r in enumerate(res):
for position,score in r:
start, end, strand = strand_adjust(position,
sizes[under20names[n]])
# Add option to round the score values. Defaulting to int atm
# since bedToBigBed only accepts integer values....
fileout[under20names[n]].write('\t'.join([chrom.name,
str(start), str(end),
under20names[n],
str(int(score*100)), strand,
'\n']))
#Run over 20s
res = MOODS.search(chrom.seq, over20, float(options.threshold),
absolute_threshold=options.A , both_strands = True,
bg=bgt, algorithm='supera')
for n,r in enumerate(res):
for position,score in r:
start, end, strand = strand_adjust(position,
sizes[over20names[n]])
fileout[over20names[n]].write('\t'.join([chrom.name, str(start),
str(end),over20names[n],
str(int(score*100)),strand,
'\n']))
def find_motif_disruptions(
position,
ref,
alt,
genome_fasta,
matrices,
):
"""
Determine whether there is a difference between the ref and alt
alleles for TF binding. Requires samtools in your path.
Parameters
----------
position : str
Zero based genomic coordinates of the reference allele of the form
chrom:start-end (chr5:100-101 for a SNV for instance). The value end -
start should equal the length of the ref allele.
ref : str
Reference allele. This should match the reference sequence at "position"
in genome_fasta.
alt : str
Alternate allele.
genome_fasta : str
Path to genome fasta file. This file should be indexed.
matrices : dict
Dict whose keys are motif names and whose values are pandas data frames
or numpy arrays containing PWMs with columns ACGT.
Returns
-------
out : pandas.DataFrame
Pandas data frame with motifs whose best matches that overlapped the
variant differed between the reference and alternate sequences. A score
of zero and a strand of '' indicates that there was not a match for the
motif on the given allele.
"""
import subprocess
import MOODS
# import pybedtools as pbt
max_motif_length = max([x.shape[0] for x in matrices.values()])
chrom, coords = position.split(':')
start,end = [int(x) for x in coords.split('-')]
s = '{}:{}-{}'.format(chrom, start - max_motif_length + 1, end +
max_motif_length - 1)
c = 'samtools faidx {} {}'.format(genome_fasta, s)
seq_lines = subprocess.check_output(c, shell=True).strip().split()
ref_seq = seq_lines[1]
alt_seq = ref_seq[0:max_motif_length - 1] + alt + ref_seq[max_motif_length + len(ref) - 1:]
ref_variant_start = max_motif_length - 1
ref_variant_end = max_motif_length - 1 + len(ref)
alt_variant_start = max_motif_length - 1
alt_variant_end = max_motif_length - 1 + len(alt)
ms = [matrices[x].T.values.tolist() for x in matrices.keys()]
ref_res = MOODS.search(ref_seq, ms, 0.001, both_strands=True,
bg=[0.25, 0.25, 0.25, 0.25])
ref_res = dict(zip(matrices.keys(), ref_res))
alt_res = MOODS.search(alt_seq, ms, 0.001, both_strands=True,
bg=[0.25, 0.25, 0.25, 0.25])
alt_res = dict(zip(matrices.keys(), alt_res))
# First we'll remove any motif matches that don't overlap the variant of interest (and thus
# can't be affected by the variant and will be the same for ref and alt). Then we'll get the
# best match for each motif for ref and alt.
rows = []
for motif in ref_res.keys():
ref_res[motif] = _filter_variant_motif_res(ref_res[motif], ref_variant_start, ref_variant_end,
matrices[motif].shape[0], ref_seq)
alt_res[motif] = _filter_variant_motif_res(alt_res[motif], alt_variant_start, alt_variant_end,
matrices[motif].shape[0], alt_seq)
if len(ref_res[motif]) > 0:
ref_pos, ref_score = sorted(ref_res[motif], key=lambda x: x[1], reverse=True)[0]
ref_strand = {True:'+', False:'-'}[ref_pos > 0]
else:
ref_score = 0
ref_strand = ''
if len(alt_res[motif]) > 0:
alt_pos, alt_score = sorted(alt_res[motif], key=lambda x: x[1], reverse=True)[0]
alt_strand = {True:'+', False:'-'}[alt_pos > 0]
else:
alt_score = 0
alt_strand = ''
if ref_score > 0 or alt_score > 0:
diff = ref_score - alt_score
rows.append([motif, ref_score, ref_strand, alt_score, alt_strand, diff])
out = pd.DataFrame(rows, columns=['motif', 'ref_score', 'ref_strand', 'alt_score',
'alt_strand', 'score_diff'])
out.index = out.motif
out = out.drop('motif', axis=1)
out = out[out.score_diff != 0]
return out
