def FRR_sam_likelihood (A, sam_path, arg_dict, weights): locus = arg_dict['locus'] chrom, locus_start, locus_end = extract_locus_info(locus) log_likelihood = 0 n = 0 with open(sam_path, 'r') as irr_handle: for record in csv.reader(irr_handle, dialect = 'excel-tab'): if record[0][0] != '@': n = n + 1 sample_dfl = extract_col_sam(record, 'om') # print sample_dfl if 'diploid' in arg_dict and arg_dict['diploid'] == 'True': samp_likelihood = weights['allele_1'] * FRR_allele_likelihood(arg_dict, A[0], sample_dfl) + \ weights['allele_2'] * FRR_allele_likelihood(arg_dict, A[1], sample_dfl) else: samp_likelihood = FRR_allele_likelihood(arg_dict, A, sample_dfl) # print record[0], '\t', locus_start - int(record[3]), '\t', samp_likelihood if samp_likelihood > 0: samp_log_likelihood = np.log(samp_likelihood) else: samp_log_likelihood = -50 log_likelihood = log_likelihood + samp_log_likelihood # log_likelihood = log_likelihood * samp_likelihood # print 'FRR count:', n return log_likelihood
def span_sam_likelihood(A, sam_path, arg_dict, weights):
locus = arg_dict['locus']
mean_ins_size = arg_dict['read_ins_mean']
chrom, locus_start, locus_end = extract_locus_info(locus)
log_likelihood = 0
yo = 0
nn = 0
with open(sam_path, 'r') as irr_handle:
for record in csv.reader(irr_handle, dialect='excel-tab'):
if record[0][0] != '@' and int(record[8]) != 0:
sample_ins = extract_col_sam(record, 'is')
if 'diploid' in arg_dict and arg_dict['diploid'] == 'True':
samp_likelihood = weights['allele_1'] * span_allele_likelihood(arg_dict, A[0], sample_ins) + \
weights['allele_2'] * span_allele_likelihood(arg_dict, A[1], sample_ins)
else:
samp_likelihood = span_allele_likelihood(
arg_dict, A, sample_ins)
# print record[0], '\t', sample_ins, '\t', np.abs(sample_ins - mean_ins_size) / 3 + 10, '\t', samp_likelihood
yo = yo + np.abs(sample_ins - mean_ins_size) / 3 + 10
nn = nn + 1
if samp_likelihood > 0:
samp_log_likelihood = np.log(samp_likelihood)
# elif np.abs(samp_likelihood) < 10**-20 and samp_likelihood != 0: # accounting for comutational errors
# print 'kek'
# samp_log_likelihood = np.log(np.abs(samp_likelihood))
elif samp_likelihood <= 0:
samp_log_likelihood = -50
log_likelihood = log_likelihood + samp_log_likelihood
# print yo / nn
return log_likelihood
def encl_sam_likelihood (A, sam_path, arg_dict, weights): locus = arg_dict['locus'] mean_ins_size = arg_dict['read_ins_mean'] chrom, locus_start, locus_end = extract_locus_info(locus) log_likelihood = 0 with open(sam_path, 'r') as irr_handle: for record in csv.reader(irr_handle, dialect = 'excel-tab'): if record[0][0] != '@': nCopy = extract_col_sam(record, 'nc') if 'diploid' in arg_dict and arg_dict['diploid'] == 'True': samp_likelihood = weights['allele_1'] * encl_allele_likelihood(arg_dict, A[0], nCopy) + \ weights['allele_2'] * encl_allele_likelihood(arg_dict, A[1], nCopy) else: samp_likelihood = encl_allele_likelihood(arg_dict, A, nCopy) # if nCopy == 30: # print record[0], '\tnC:', nCopy, '\t', samp_likelihood if samp_likelihood > 0: samp_log_likelihood = np.log(samp_likelihood) # elif np.abs(samp_likelihood) < 10**-20 and samp_likelihood != 0: # accounting for comutational errors # print samp_likelihood, A, nCopy # samp_log_likelihood = np.log(np.abs(samp_likelihood)) elif samp_likelihood <= 0: samp_log_likelihood = -50 log_likelihood = log_likelihood + samp_log_likelihood # print yo / nn return log_likelihood
def IRR_sam_likelihood(A, B, sam_path, arg_dict):
locus = arg_dict['locus']
chrom, locus_start, locus_end = extract_locus_info(locus)
log_likelihood = 1
with open(sam_path, 'r') as irr_handle:
for record in csv.reader(irr_handle, dialect='excel-tab'):
if record[0][0] != '@':
sample_dfl = locus_start - int(record[3])
samp_likelihood = IRR_genotype_likelihood(
arg_dict, A, B, sample_dfl)
# print record[0], '\t', locus_start - int(record[3]), '\t', samp_likelihood
if samp_likelihood > 0:
samp_log_likelihood = np.log(samp_likelihood)
else:
samp_log_likelihood = -250
log_likelihood = log_likelihood + samp_log_likelihood
# log_likelihood = log_likelihood * samp_likelihood
return log_likelihood
def span_sam_likelihood(A, B, sam_path, arg_dict):
locus = arg_dict['locus']
chrom, locus_start, locus_end = extract_locus_info(locus)
log_likelihood = 0
with open(sam_path, 'r') as irr_handle:
for record in csv.reader(irr_handle, dialect='excel-tab'):
if record[0][0] != '@' and int(record[8]) != 0:
sample_ins = int(record[8])
samp_likelihood = span_genotype_likelihood(
arg_dict, A, B, sample_ins)
# print record[0], '\t', int(record[8]), '\t', np.abs(int(record[8]) - 500) / 3 + 10, '\t', samp_likelihood
if samp_likelihood > 0:
samp_log_likelihood = np.log(samp_likelihood)
elif np.abs(samp_likelihood
) < 10**-20: # accounting for comutational errors
samp_log_likelihood = np.log(np.abs(samp_likelihood))
elif samp_likelihood == 0:
samp_log_likelihood = -250
else:
print 'Error! Negative likelihood:', samp_likelihood
log_likelihood = log_likelihood + samp_log_likelihood
return log_likelihood
parser.add_argument('--in-pref', type=str, required=True)
parser.add_argument('--exp-dir', type=str, required=True)
args = parser.parse_args()
out_pref = args.out_pref
in_pref = args.in_pref
exp_dir = args.exp_dir
arg_dict = load_profile(exp_dir)
read_len = arg_dict['read_len']
locus = arg_dict['locus']
motif = arg_dict['motif']
chrom, locus_start, locus_end = extract_locus_info(locus)
pre, post = extract_pre_post_flank(exp_dir, read_len)
score_dict = { 'match': 3, \
'mismatch': -1, \
'gap': -3}
verbose = False
margin = 2
in_sam = in_pref + '.sam'
out_sam = out_pref + '.sam'
out_sam_handle = open(out_sam, 'w')
print 'Filtering ' + in_pref + '.sam'
with open(in_sam, 'r') as in_sam_handle:
for record in in_sam_handle:
if record[0] == '@':
from realignment import expansion_aware_realign, classify_realigned_read
from load_info import load_profile, extract_locus_info
from extract_genome import extract_pre_post_flank
read_class = 'srp'
nCopy = 70
filt_path = '/storage/nmmsv/expansion-experiments/ATXN3_32_cov60_dist500_hap_viz/aligned_read/nc_'+str(nCopy)+'_'+read_class+'.sam'
# filt_path = '/storage/nmmsv/python_playground/test_filter_IRR/nc_'+str(nCopy)+'.sam'
filt_path_true = '/storage/nmmsv/expansion-experiments/ATXN3_32_cov60_dist500_hap_viz/aligned_read/true_filter/nc_'+str(nCopy)+'_'+read_class+'.sam'
sam_path = '/storage/nmmsv/expansion-experiments/ATXN3_32_cov60_dist500_hap_viz/aligned_read/nc_'+str(nCopy)+'.sam'
exp_dir = '/storage/nmmsv/expansion-experiments/ATXN3_32_cov60_dist500_hap_viz/'
arg_dict = load_profile(exp_dir)
locus = arg_dict['locus']
read_len = arg_dict['read_len']
motif = arg_dict['motif']
chrom, locus_start_ref, locus_end_ref = extract_locus_info(locus)
pre, post = extract_pre_post_flank(exp_dir, read_len)
score_dict = { 'match': 3, \
'mismatch': -1, \
'gap': -3}
verbose = False
margin = 2
print locus_start_ref, locus_end_ref
true_reads = []
kk = 0
with open (filt_path, 'r') as handle:
for record in handle:
if record[0] != '@':
kk = kk + 1
QNAME = record.split()[0]