def leven_search_pam(pam_obj_file, target, threshold, out_file):
target = str.upper(target)
targ_len = len(target)
d = 0
hit = []
with open(pam_obj_file, 'rb') as pam_handle:
with open(out_file, 'wb') as csvfile:
pam_obj = pickle.load(pam_handle)
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
ct = 0
for el in pam_obj:
d = levenshtein(el[0].upper(), target)
if d <= threshold:
hit.append(el + [d])
ct += 1
if (ct % 1000) == 0:
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
print str(float(ct) / float(len(pam_obj)))
mywriter = csv.writer(csvfile, delimiter=',')
for el in hit:
mywriter.writerow(el)
def nt_query_gg_all(gg_obj_file, out_file):
with open(gg_obj_file, 'rb') as gg_handle:
gg_obj = pickle.load(gg_handle)
rf_split = gg_obj_file.split('_')
chrid = rf_split[len(rf_split) - 2]
offset = int(rf_split[len(rf_split) - 1].replace('.gg', ''))
print offset
with open(out_file, 'wb') as csvfile:
mywriter = csv.writer(csvfile, delimiter=',')
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
#skip overlap if offset > 0
if offset == 0:
ref_lb = 0
else:
ref_lb = 1000
ref_ub = len(gg_obj.ref)
out = gg_obj.get_ntvar_counts(ref_lb, ref_ub)
mywriter.writerow([chrid, offset] + out['A'] + out['C'] +
out['G'] + out['T'])
def nt_query_gg_all(gg_obj_file,gff3_file,out_file):
with open(gg_obj_file,'rb') as gg_handle:
gg_obj = pickle.load(gg_handle)
with open(out_file,'wb') as csvfile:
mywriter = csv.writer(csvfile, delimiter=',')
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
gff3_file_list = gff3_file.replace('.gff3','').split('_')
offset = int(gff3_file_list[len(gff3_file_list)-1])
with open(gff3_file,'rb') as fin:
for line in fin:
if line[0] == '#':
continue
line_els = line.strip().split()
if line_els[2] == 'exon':
ref_lb = int(line_els[3])-offset-1
ref_ub = int(line_els[4])-offset-1
if ref_lb < 0:
ref_lb = 0
if ref_ub > len(gg_obj.ref):
ref_ub = len(gg_obj.ref)
out = gg_obj.get_ntvar_counts(ref_lb,ref_ub)
mywriter.writerow(line_els+out['A']+out['T']+out['C']+out['G']+out['Cnot']+out['Gnot'])
def __init__(self, geno_genome):
super(Geno_slice, self).__init__()
self.ref = geno_genome.ref
self.var_track = geno_genome.var_track
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
self.ind = 0
self.len = 0
self.name_dict = {}
self.geno_dict = {}
def split_chrom_ref_vcf_gff3(ref_filename,vcf_filename,gff3_filename):
with open(ref_filename,'r') as fin:
for record in SeqIO.parse(fin, "fasta"):
print record.id
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
with open(ref_filename.replace('.fa',
'_'+str(record.id)+".fa"),'wb') as fout:
SeqIO.write(record, fout, "fasta")
def leven_search_pam(pam_obj_file,target_file,threshold,out_file):
pof_base = os.path.basename(pam_obj_file)
with open(pam_obj_file,'rb') as pam_handle:
pam_obj = pickle.load(pam_handle)
with open(out_file,'wb') as csvfile:
mywriter = csv.writer(csvfile, delimiter=',')
with open(target_file,'rb') as targfile:
for record in SeqIO.parse(targfile, "fasta"):
target = str(record.seq).upper()
name_list = str(record.name).split('|')
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
ct = 0
d = 0
hit = []
for el in pam_obj:
# d = levenshtein(el[0].upper(),target)
d = linear(el[0].upper(),target)
if d <= threshold:
hit.append(name_list+[target,pof_base,d]+el)
ct+=1
if (ct%100000) == 0:
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
print str(float(ct)/float(len(pam_obj)))
for el in hit:
mywriter.writerow(el)
def target_query_gg_all(gg_obj_file, gff3_file, PAM_orient, PAM_list,
target_len, out_file):
PAM_len = len(PAM_list[0])
PAM_dict = {}
for pam in PAM_list:
PAM_dict[str(pam)] = pam
PAM_dict_rc = {}
for pam in PAM_list:
PAM_dict_rc[revcomp(str(pam))] = pam
gff3_pct = {}
gff3_file_list = gff3_file.replace('.gff3', '').split('_')
offset = int(gff3_file_list[len(gff3_file_list) - 1])
with open(gff3_file, 'rb') as fin:
for line in fin:
if line[0] == '#':
continue
row_obj = gff3.GFF3_row(line)
# print str(row_obj)
#protein coding transcript (PCT) and exon or UTR (EU)
if (row_obj.PCT and row_obj.EU):
if row_obj.INFO['transcript_id'] not in gff3_pct:
gff3_pct[row_obj.INFO['transcript_id']] = gff3.PCT(row_obj)
else:
gff3_pct[row_obj.INFO['transcript_id']].add_element(
row_obj)
gff3_lines = []
for pct_id in gff3_pct:
gff3_lines += gff3_pct[pct_id].get_coding()
# print str(gff3_pct[pct_id])
# print gff3_lines
with open(gg_obj_file, 'rb') as gg_handle:
gcr_obj = gg.Geno_CRISPR(pickle.load(gg_handle))
with open(out_file, 'wb') as csvfile:
mywriter = csv.writer(csvfile, delimiter=',')
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
for line_els in gff3_lines:
# if line[0] == '#':
# continue
# line_els = line.strip().split()
if line_els[2] == 'exon':
ref_lb = int(line_els[3]) - offset - 1
ref_ub = int(line_els[4]) - offset - 1
if ref_lb < 0:
ref_lb = 0
if ref_ub > gcr_obj.len:
ref_ub = gcr_obj.len
[target_inds,
target_vars] = gcr_obj.get_var_targets_del_pams(
PAM_orient, PAM_list, target_len, ref_lb, ref_ub)
n_targ = len(target_inds)
n_var = len(target_vars)
for loc in target_inds:
af = 0
hetf = 0
max_var = 0
max_var_els = []
#change 161006 taking max only
if str(loc) in target_vars:
# print [[el[0],el[1].get_af_adj()] for el in target_vars[str(loc)]]
for el in target_vars[str(loc)]:
if el[1].get_af_adj() > max_var:
max_var = el[1].get_af_adj()
max_var_els = [el]
elif el[1].get_af_adj() == max_var:
max_var_els += [el]
#pam start is index 0 at this point
#for equal af, take closest to PAM
max_var_els.sort(key=lambda x: x[0])
ind = 0
el = max_var_els[0]
if PAM_orient == 'R':
#orient PAM at right hand side
ind = (target_len + PAM_len) - (el[0] + 1)
elif PAM_orient == 'L':
ind = el[0]
af = el[1].get_af_adj()
hetf = el[1].get_hetf()
#platinum: af < 0.0001
if af < 0.0001:
out_list = [
loc[0], loc[0] + len(loc[2]), loc[1],
PAM_orient
]
if PAM_orient == 'R':
#orient PAM at right hand side
out_list += [loc[2][0:target_len]]
out_list += [loc[2][target_len:len(loc[2])]]
elif PAM_orient == 'L':
out_list += [loc[2][PAM_len:len(loc[2])]]
out_list += [loc[2][0:PAM_len]]
out_list += [af]
mywriter.writerow(line_els + out_list)
def PAM_query_gg_all(gg_obj_file, gff3_file, PAM_list, out_file):
PAM_len = len(PAM_list[0])
PAM_dict = {}
for pam in PAM_list:
PAM_dict[str(pam)] = pam
PAM_dict_rc = {}
for pam in PAM_list:
PAM_dict_rc[revcomp(str(pam))] = pam
gff3_pct = {}
gff3_file_list = gff3_file.replace('.gff3', '').split('_')
offset = int(gff3_file_list[len(gff3_file_list) - 1])
with open(gff3_file, 'rb') as fin:
for line in fin:
if line[0] == '#':
continue
row_obj = gff3.GFF3_row(line)
# print str(row_obj)
#protein coding transcript (PCT) and exon or UTR (EU)
if (row_obj.PCT and row_obj.EU):
if row_obj.INFO['transcript_id'] not in gff3_pct:
gff3_pct[row_obj.INFO['transcript_id']] = gff3.PCT(row_obj)
else:
gff3_pct[row_obj.INFO['transcript_id']].add_element(
row_obj)
gff3_lines = []
for pct_id in gff3_pct:
gff3_lines += gff3_pct[pct_id].get_coding()
# print str(gff3_pct[pct_id])
# print gff3_lines
with open(gg_obj_file, 'rb') as gg_handle:
gcr_obj = gg.Geno_CRISPR(pickle.load(gg_handle))
with open(out_file, 'wb') as csvfile:
mywriter = csv.writer(csvfile, delimiter=',')
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
for line_els in gff3_lines:
# if line[0] == '#':
# continue
# line_els = line.strip().split()
if line_els[2] == 'exon':
ref_lb = int(line_els[3]) - offset - 1
ref_ub = int(line_els[4]) - offset - 1
if ref_lb < 0:
ref_lb = 0
if ref_ub > gcr_obj.len:
ref_ub = gcr_obj.len
af_bins = [[0.00001, 0], [0.0001, 0], [0.001, 0],
[0.01, 0], [0.1, 0], [1, 0]]
hetf_bins = [[0.00001, 0], [0.0001, 0], [0.001, 0],
[0.01, 0], [0.1, 0], [1, 0]]
[pam_inds,
del_vars] = gcr_obj.get_del_PAMS(PAM_list, ref_lb, ref_ub)
n_pam = len(pam_inds)
n_del = len(del_vars)
for loc in del_vars:
af = 0
hetf = 0
for var in del_vars[loc]:
#change 161006 taking max only
if var.get_af_adj() > af:
af = var.get_af_adj()
hetf = var.get_hetf()
#change 161006 taking max only
varct = 1
if af < af_bins[0][0]:
af_bins[0][1] += 1
hetf_bins[0][1] += hetf / varct
elif af < af_bins[1][0]:
af_bins[1][1] += 1
hetf_bins[1][1] += hetf / varct
elif af < af_bins[2][0]:
af_bins[2][1] += 1
hetf_bins[2][1] += hetf / varct
elif af < af_bins[3][0]:
af_bins[3][1] += 1
hetf_bins[3][1] += hetf / varct
elif af < af_bins[4][0]:
af_bins[4][1] += 1
hetf_bins[4][1] += hetf / varct
elif af <= af_bins[5][0]:
af_bins[5][1] += 1
hetf_bins[5][1] += hetf / varct
mywriter.writerow(line_els + [el[1] for el in af_bins] +
[el[1] for el in hetf_bins] +
[n_del, n_pam, 'del'])
af_bins = [[0.00001, 0], [0.0001, 0], [0.001, 0],
[0.01, 0], [0.1, 0], [1, 0]]
hetf_bins = [[0.00001, 0], [0.0001, 0], [0.001, 0],
[0.01, 0], [0.1, 0], [1, 0]]
[no_pam_inds,
add_vars] = gcr_obj.get_add_PAMS(PAM_list, ref_lb, ref_ub)
n_no_pam = len(no_pam_inds)
n_add = len(add_vars)
for loc in add_vars:
af = 0
hetf = 0
for var in add_vars[loc]:
#change 161006 taking max only
if var.get_af_adj() > af:
af = var.get_af_adj()
hetf = var.get_hetf()
#change 161006 taking max only
varct = 1
if af < af_bins[0][0]:
af_bins[0][1] += 1
hetf_bins[0][1] += hetf / varct
elif af < af_bins[1][0]:
af_bins[1][1] += 1
hetf_bins[1][1] += hetf / varct
elif af < af_bins[2][0]:
af_bins[2][1] += 1
hetf_bins[2][1] += hetf / varct
elif af < af_bins[3][0]:
af_bins[3][1] += 1
hetf_bins[3][1] += hetf / varct
elif af < af_bins[4][0]:
af_bins[4][1] += 1
hetf_bins[4][1] += hetf / varct
elif af <= af_bins[5][0]:
af_bins[5][1] += 1
hetf_bins[5][1] += hetf / varct
mywriter.writerow(line_els + [el[1] for el in af_bins] +
[el[1] for el in hetf_bins] +
[n_add, n_no_pam, 'add'])
import vmb print '####################' print 'vmb memory:' + str(vmb.memory()) print 'vmb resident: ' + str(vmb.resident()) print '####################'
def PAM_compile_gg(gg_obj_file, targ_len, PAM_orient, PAM_list, out_file):
PAM_len = len(PAM_list[0])
PAM_dict = {}
for pam in PAM_list:
PAM_dict[str(pam)] = pam
PAM_dict_rc = {}
for pam in PAM_list:
PAM_dict_rc[revcomp(str(pam))] = pam
with open(gg_obj_file, 'rb') as gg_handle:
with open(out_file, 'wb') as fout:
ggs_obj = gg.Geno_slice(pickle.load(gg_handle))
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
print ggs_obj.ref
# print ggs_obj.var_track
len_slice = targ_len + (2 * PAM_len)
ggs_obj.slice(0, len_slice)
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
ct = 0
ct2 = 0
search_seqs = []
for i in range(0, (len(ggs_obj.ref) - len_slice)):
gg_seqs = ggs_obj.compile_geno_seqs()
gg_seqs += [
gg.Geno_seq(ggs_obj.ref[i:(i + len_slice)],
ggs_obj.ind - ggs_obj.len, -1, ['0'])
]
for gg_seq in gg_seqs:
if PAM_orient == 'R':
if gg_seq.seq[0:PAM_len] in PAM_dict_rc:
# if (levenshtein(revcomp(gg_seq.seq[PAM_len:(PAM_len+targ_len)]),
# target) <= threshold):
# fout.write('>'+gg_seq.seq[0:PAM_len]+
# '|'+PAM_orient+'|'+'BS'+'|'+str(gg_seq)+'\n')
# fout.write(revcomp(gg_seq.seq[PAM_len:(PAM_len+targ_len)])+'\n')
search_seqs.append([
revcomp(gg_seq.seq[PAM_len:(PAM_len +
targ_len)]),
gg_seq.seq[0:PAM_len], PAM_orient, 'BS',
gg_seq.ind, gg_seq.allele, gg_seq.names
])
ct2 += 1
if gg_seq.seq[(len_slice -
PAM_len):len_slice] in PAM_dict:
# if (levenshtein(gg_seq.seq[PAM_len:(PAM_len+targ_len)],
# target) <= threshold):
# fout.write('>'+gg_seq.seq[(len_slice-PAM_len):len_slice]+
# '|'+PAM_orient+'|'+'TS'+'|'+str(gg_seq)+'\n')
# fout.write(gg_seq.seq[PAM_len:(PAM_len+targ_len)]+'\n')
search_seqs.append([
gg_seq.seq[PAM_len:(PAM_len + targ_len)],
gg_seq.seq[(len_slice - PAM_len):len_slice],
PAM_orient, 'TS', gg_seq.ind, gg_seq.allele,
gg_seq.names
])
ct2 += 1
elif PAM_orient == 'L':
if gg_seq.seq[0:PAM_len] in PAM_dict:
# if (levenshtein(gg_seq.seq[PAM_len:(PAM_len+targ_len)],
# target) <= threshold):
# fout.write('>'+gg_seq.seq[0:PAM_len]+
# '|'+PAM_orient+'|'+'TS'+'|'+str(gg_seq)+'\n')
# fout.write(gg_seq.seq[PAM_len:(PAM_len+targ_len)]+'\n')
search_seqs.append([
gg_seq.seq[PAM_len:(PAM_len + targ_len)],
gg_seq.seq[0:PAM_len], PAM_orient, 'TS',
gg_seq.ind, gg_seq.allele, gg_seq.names
])
ct2 += 1
if gg_seq.seq[(len_slice -
PAM_len):len_slice] in PAM_dict_rc:
# if (levenshtein(revcomp(gg_seq.seq[PAM_len:(PAM_len+targ_len)]),
# target) <= threshold):
# fout.write('>'+gg_seq.seq[(len_slice-PAM_len):len_slice]+
# '|'+PAM_orient+'|'+'BS'+'|'+str(gg_seq)+'\n')
# fout.write(revcomp(gg_seq.seq[PAM_len:(PAM_len+targ_len)])+'\n')
search_seqs.append([
revcomp(gg_seq.seq[PAM_len:(PAM_len +
targ_len)]),
gg_seq.seq[(len_slice - PAM_len):len_slice],
PAM_orient, 'BS', gg_seq.ind, gg_seq.allele,
gg_seq.names
])
ct2 += 1
ct += 1
ggs_obj.ref_base_shift()
if (i % 100000) == 0:
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
print str(
float(i) /
(float(len(ggs_obj.ref)) - float(len_slice)))
print ct2
pickle.dump(search_seqs, fout)
def gene_query_gg_all(gg_obj_file, gff3_file, genes_file, PAM_orient, PAM_list,
target_len, out_file):
PAM_len = len(PAM_list[0])
PAM_dict = {}
for pam in PAM_list:
PAM_dict[str(pam)] = pam
PAM_dict_rc = {}
for pam in PAM_list:
PAM_dict_rc[revcomp(str(pam))] = pam
genes = []
with open(genes_file, 'rb') as csvgene:
greader = csv.reader(csvgene, delimiter=',')
genes = [row for row in greader]
print genes
gff3_pct = {}
gff3_file_list = gff3_file.replace('.gff3', '').split('_')
offset = int(gff3_file_list[len(gff3_file_list) - 1])
with open(gff3_file, 'rb') as fin:
for line in fin:
if line[0] == '#':
continue
row_obj = gff3.GFF3_row(line)
# print str(row_obj)
#protein coding transcript (PCT) and exon or UTR (EU)
if (row_obj.PCT and row_obj.EU):
if row_obj.INFO['transcript_id'] not in gff3_pct:
gff3_pct[row_obj.INFO['transcript_id']] = gff3.PCT(row_obj)
else:
gff3_pct[row_obj.INFO['transcript_id']].add_element(
row_obj)
gff3_lines = []
for pct_id in gff3_pct:
gff3_lines += gff3_pct[pct_id].get_coding()
# print str(gff3_pct[pct_id])
# print gff3_lines
with open(gg_obj_file, 'rb') as gg_handle:
gcr_obj = gg.Geno_CRISPR(pickle.load(gg_handle))
with open(out_file, 'wb') as csvfile:
mywriter = csv.writer(csvfile, delimiter=',')
print '####################'
print 'vmb memory:' + str(vmb.memory())
print 'vmb resident: ' + str(vmb.resident())
print '####################'
for gene in genes:
gene_id = gene[0]
print gene_id
for line_els in gff3_lines:
# if line[0] == '#':
# continue
# line_els = line.strip().split()
if ((line_els[2] == 'exon') and (gene_id in line_els[1])):
ref_lb = int(line_els[3]) - offset - 1
ref_ub = int(line_els[4]) - offset - 1
if ref_lb < 0:
ref_lb = 0
if ref_ub > gcr_obj.len:
ref_ub = gcr_obj.len
[target_inds,
target_vars] = gcr_obj.get_var_targets_del_pams(
PAM_orient, PAM_list, target_len, ref_lb, ref_ub)
n_targ = len(target_inds)
n_var = len(target_vars)
# print target_inds
# print target_vars
for loc in target_inds:
all_freq = numpy.zeros(target_len +
PAM_len).astype(float)
het_freq = numpy.zeros(target_len +
PAM_len).astype(float)
hom_freq = numpy.zeros(target_len +
PAM_len).astype(float)
ind_dict = {}
if str(loc) in target_vars:
for el in target_vars[str(loc)]:
if PAM_orient == 'R':
#orient PAM at right hand side
ind = (target_len + PAM_len) - (el[0] +
1)
elif PAM_orient == 'L':
ind = el[0]
#take max af allele for each position
#then max of all_freq is max af of target
if ind not in ind_dict:
ind_dict[ind] = el
else:
if el[1].get_af_adj(
) > ind_dict[ind][1].get_af_adj():
ind_dict[ind] = el
for ind in ind_dict:
all_freq[ind] = ind_dict[ind][
1].get_af_adj()
het_freq[ind] = ind_dict[ind][
1].get_af_het_adj()
hom_freq[ind] = ind_dict[ind][
1].get_af_hom_adj()
loc[0] = loc[0] + offset
mywriter.writerow(line_els + list(all_freq) + loc +
[n_var, n_targ, 'ALL'] + gene)
mywriter.writerow(line_els + list(het_freq) + loc +
[n_var, n_targ, 'HET'] + gene)
mywriter.writerow(line_els + list(hom_freq) + loc +
[n_var, n_targ, 'HOM'] + gene)