def make_tsvfile_predictions(
namefile,
modfile,
ucsc_exe,
ucsc_dbs,
web=False,
win=2,
dbfasta='hg38.2bit',
dbpps=['hg38.phyloP7way.bw', 'hg38.phyloP100way.bw'],
pklcod='hg38_coding.pkl',
fprog='twoBitToFa',
cprog='bigWigToBedGraph'):
nucs = 'ACGTN'
try:
model1 = joblib.load(modfile[0])
model2 = joblib.load(modfile[1])
except:
print >> sys.stderr, 'ERROR: Program not able to load modfile. Please check that you have installed a compatible version joblib.'
sys.exit(1)
proc = subprocess.Popen([prog_cat, '-f', namefile],
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
stdout, stderr = proc.communicate()
c = 0
print "#CHROM\tPOS\tREF\tALT\tCODING\tPREDICTION\tSCORE\tFDR\tPhyloP100\tAvgPhyloP100"
for line in stdout.split('\n'):
if line == '': continue
v = line.rstrip().split()
c = c + 1
if len(v) < 4:
print >> sys.stderr, 'ERROR: Incorrect line ', c
print >> sys.stderr, 'ERROR:', line
continue
(ichr, pos, wt, nw) = tuple(v[:4])
if wt == nw or nw.find(',') > -1:
print >> sys.stderr, 'ERROR: Incorrect input line.', ichr, pos, wt, nw
print >> sys.stderr, 'ERROR:', line
continue
nchr = ichr
if nchr.find('chr') == -1: nchr = 'chr' + ichr
try:
ipos = int(pos)
except:
print >> sys.stderr, 'ERROR: Incorrect input data. The tsv input file should have has four columns (chr,position,ref,alt).'
print >> sys.stderr, 'ERROR:', line
continue
sys.exit(1)
lwt = len(wt)
lnw = len(nw)
##Option for selecting snv
#if lwt!=1 or lnw!=1 or nucs.find(wt)==-1 or nucs.find(nw)==-1 or wt==nw:
# print >> sys.stderr, 'ERROR: Not single nucloetide variant',','.join([nchr,pos,wt,nw])
# continue
if wt == '-':
lwt = 1
lnw += 1
if nw == '-':
lwt += 1
lnw = 1
n_wt, n_nw, n_pos = parse_variants(nchr, ipos, wt, nw, ucsc_exe,
ucsc_dbs, web, dbfasta, fprog)
if n_wt == '' or n_nw == '':
print >> sys.stderr, 'ERROR: Incorrect mutation mapping. Check position', ichr, ipos, wt, nw
print >> sys.stderr, 'ERROR:', line
continue
if 'ACGTN'.find(n_wt) == -1 or 'ACGTN'.find(n_nw) == -1:
print >> sys.stderr, 'ERROR: Incorrect wild-type or mutant nucleotide', wt, nw
print >> sys.stderr, 'ERROR:', line
continue
if len(wt) == 1 and len(nw) == 1:
r_cod = []
(nuc, seq, seq_input, cons_input,
r_cod) = get_snv_input(nchr, n_pos, n_wt, n_nw, ucsc_exe,
ucsc_dbs, web, win, dbfasta, dbpps, pklcod,
fprog, cprog)
else:
(nuc, seq, seq_input, cons_input,
r_cod) = get_indel_input(nchr, n_pos, n_wt, n_nw, ucsc_exe,
ucsc_dbs, web, win, dbfasta, dbpps,
pklcod, fprog, cprog)
if seq == '':
print >> sys.stderr, 'ERROR: Sequence not found for line ' + str(
c) + '. Genome location:', ichr, pos
print >> sys.stderr, 'ERROR:', line
continue
if seq_input == []:
print >> sys.stderr, 'ERROR: Incorrect nucleotide in line ' + str(
c) + '. Genome location:', ichr, pos
print >> sys.stderr, 'ERROR:', line
continue
if cons_input != []:
cons_input1 = cons_input[0]
cons_input2 = cons_input[1]
else:
print >> sys.stderr, 'ERROR: Incorrect conservation data for line ' + str(
c) + '. Genome location:', ichr, pos
print >> sys.stderr, 'ERROR:', line
continue
#if cons_input1==[] or cons_input2==[]:
#Check only P100
if cons_input2 == []:
print >> sys.stderr, 'ERROR: Incorrect conservation data for line ' + str(
c) + '. Genome location:', ichr, pos
print >> sys.stderr, 'ERROR:', line
continue
if cons_input1 == []:
cons_input1 = [0.0 for i in range(2 * win + 1)]
print >> sys.stderr, 'WARNING: PhyloP7 data not found for line', c, 'mutated site', ichr, pos
p_cod = 0
cod = 'No'
if r_cod != []:
p_cod = 1
cod = 'Yes'
if len(wt) == 1 and len(nw) == 1 and wt != '-' and nw != '-':
X = [seq_input + cons_input1 + cons_input2]
y_pred, y_fdrs, c_pred = prediction(X, model1)
v_fdr = [y_fdrs[0][0], y_fdrs[0][1]]
else:
X = [seq_input + cons_input1 + cons_input2 + [lwt, lnw, p_cod]]
y_pred, y_fdrs, c_pred = prediction(X, model2)
v_fdr = [y_fdrs[0][2], y_fdrs[0][3]]
if y_pred == []:
print >> sys.stderr, 'WARNING: Variant not scored. Check modfile and input'
print >> sys.stderr, 'WARNING:', line
continue
pp100 = cons_input2[win]
avgpp100 = sum(cons_input2) / float(len(cons_input2))
#print pp100,avgpp100,cons_input2
if c_pred[0] == "Pathogenic": d_fdr = v_fdr[0]
if c_pred[0] == "Benign": d_fdr = v_fdr[1]
print line + '\t' + '%s\t%s\t%.3f\t%.3f\t%.3f\t%.3f' % (
cod, c_pred[0], y_pred[0], d_fdr, pp100, avgpp100)
#print '\t'.join(str(i) for i in [ichr,ipos,wt,nw,'%.4f' %y_pred[0]])
return
def make_file_predictions(namefile,
modfile,
ucsc_exe,
ucsc_dbs,
web=False,
win=2,
s='\t',
dbfasta='hg38.2bit',
dbpps=['hg38.phyloP7way.bw', 'hg38.phyloP100way.bw'],
pklcod='hg38_coding.pkl',
fprog='twoBitToFa',
cprog='bigWigToBedGraph'):
try:
model1 = joblib.load(modfile[0])
model2 = joblib.load(modfile[1])
except:
print >> sys.stderr, 'ERROR: Program not able to load modfile. Please check that you have installed a compatible version joblib.'
sys.exit(1)
f = open(namefile)
c = 1
print "#CHROM\tPOS\tREF\tALT\tCODING\tPREDICTION\tSCORE\tFDR\tPhyloP100\tAvgPhyloP100"
for line in f:
v = line.rstrip().split(s)
if len(v) < 4:
print >> sys.stderr, 'ERROR: Incorrect line ', c, line.rstrip()
print >> sys.stderr, line
continue
(ichr, pos, wt, nw) = v[:4]
try:
ipos = int(pos)
except:
print >> sys.stderr, 'ERROR: Incorrect genome location', pos, '. Check your input file'
print >> sys.stderr, 'ERROR:', line
continue
lwt = len(wt)
lnw = len(nw)
if wt == '-':
lwt = 1
lnw += 1
if nw == '-':
lwt += 1
lnw = 1
n_wt, n_nw, n_pos = parse_variants(ochr, ipos, wt, nw, ucsc_exe,
ucsc_dbs, web, dbfasta, fprog)
if n_wt == '' or n_nw == '':
print >> sys.stderr, 'ERROR: Incorrect mutation mapping. Check position', ichr, ipos, wt, nw
print >> sys.stderr, 'ERROR:', line
continue
if 'ACGTN'.find(n_wt) == -1 or 'ACGTN'.find(n_nw) == -1:
print >> sys.stderr, 'ERROR: Incorrect wild-type or mutant nucleotide', wt, nw
print >> sys.stderr, 'ERROR:', line
continue
if wt == nw or nw.find(',') > -1:
print >> sys.stderr, 'ERROR: Incorrect wild-type or mutant nucleotide', wt, nw
print >> sys.stderr, 'ERROR:', line
continue
if len(wt) == 1 and len(nw) == 1:
r_cod = []
(nuc, seq, seq_input, cons_input,
r_cod) = get_snv_input(ichr, n_pos, n_wt, n_nw, ucsc_exe,
ucsc_dbs, web, win, dbfasta, dbpps, pklcod,
fprog, cprog)
else:
(nuc, seq, seq_input, cons_input,
r_cod) = get_indel_input(ichr, n_pos, n_wt, n_nw, ucsc_exe,
ucsc_dbs, web, win, dbfasta, dbpps,
pklcod, fprog, cprog)
if seq == '':
print >> sys.stderr, 'ERROR: Sequence not found for line', c, ichr, pos
print >> sys.stderr, 'ERROR:', line
continue
if seq_input == []:
print >> sys.stderr, 'ERROR: Incorrect nucleotide in line ' + str(
c) + '. Genome location:', ichr, pos
print >> sys.stderr, 'ERROR:', line
continue
if cons_input != []:
cons_input1 = cons_input[0]
cons_input2 = cons_input[1]
else:
print >> sys.stderr, 'ERROR: Incorrect conservation data for line', c, ichr, pos
print >> sys.stderr, 'ERROR:', line
continue
if cons_input2 == []:
print >> sys.stderr, 'ERROR: Incorrect conservation data for line', c, ichr, pos
print >> sys.stderr, 'ERROR:', line
continue
if cons_input1 == []:
cons_input1 = [0.0 for i in range(2 * win + 1)]
print >> sys.stderr, 'WARNING: PhyloP7 data not found for line', c, ichr, pos
p_cod = 0
cod = 'No'
if r_cod != []:
p_cod = 1
cod = 'Yes'
if len(wt) == 1 and len(nw) == 1 and wt != '-' and nw != '-':
X = [seq_input + cons_input1 + cons_input2]
y_pred, y_fdrs, c_pred = prediction(X, model1)
v_fdr = [y_fdrs[0][0], y_fdrs[0][1]]
else:
X = [seq_input + cons_input1 + cons_input2 + [lwt, lnw, p_cod]]
y_pred, y_fdrs, c_pred = prediction(X, model2)
v_fdr = [y_fdrs[0][2], y_fdrs[0][3]]
if y_pred == []:
print >> sys.stderr, 'WARNING: Variant not scored. Check modfile and input'
print >> sys.stderr, 'WARNING:', line
continue
pp100 = cons_input2[win]
avgpp100 = sum(cons_input2) / float(len(cons_input2))
if c_pred[0] == "Pathogenic": d_fdr = v_fdr[0]
if c_pred[0] == "Benign": d_fdr = v_fdr[1]
print '\t'.join(
str(i) for i in [
ichr, ipos, wt, nw, cod, c_pred[0],
'%.3f' % y_pred[0],
'%.3f' % d_fdr, pp100, avgpp100
])
return
def make_prediction(ichr,
ipos,
wt,
nw,
modfile,
ucsc_exe,
ucsc_dbs,
web=False,
win=2,
dbfasta='hg38.2bit',
dbpps=['hg38.phyloP7way.bw', 'hg38.phyloP100way.bw'],
pklcod='',
fprog='twoBitToFa',
cprog='bigWigToBedGraph'):
lwt = len(wt)
lnw = len(nw)
if wt == '-':
lwt = 1
lnw += 1
if nw == '-':
lwt += 1
lnw = 1
n_wt, n_nw, n_pos = parse_variants(ochr, ipos, wt, nw, ucsc_exe, ucsc_dbs,
web, dbfasta, fprog)
if n_wt == '' or n_nw == '':
print >> sys.stderr, 'ERROR: Incorrect mutation mapping. Check position', ichr, ipos, wt, nw
sys.exit(1)
if 'ACGTN'.find(n_wt) == -1 or 'ACGTN'.find(n_nw) == -1:
print >> sys.stderr, 'ERROR: Incorrect wild-type or mutant nucleotide', wt, nw
sys.exit(1)
if pklcod == '':
(nuc, seq, seq_input, cons_input,
r_cod) = get_snv_input(ichr, n_pos, n_wt, n_nw, ucsc_exe, ucsc_dbs,
web, win, dbfasta, dbpps, pklcod, fprog, cprog)
else:
(nuc, seq, seq_input, cons_input,
r_cod) = get_indel_input(ichr, n_pos, n_wt, n_nw, ucsc_exe, ucsc_dbs,
web, win, dbfasta, dbpps, pklcod, fprog,
cprog)
if seq == '':
print >> sys.stderr, 'ERROR: Sequence not found for position', ichr, ipos
sys.exit(1)
if seq_input == []:
print >> sys.stderr, 'ERROR: Incorrect nucleotide in position', ichr, ipos
sys.exit(1)
if cons_input.count([]) > 0:
print >> sys.stderr, 'ERROR: Incorrect conservation data in position', ichr, ipos
sys.exit(1)
if cons_input != []:
cons_input1 = cons_input[0]
cons_input2 = cons_input[1]
else:
print >> sys.stderr, 'ERROR: Incorrect conservation data for position', ichr, ipos
sys.exit(1)
if cons_input2 == []:
print >> sys.stderr, 'ERROR: Incorrect conservation data for position', ichr, pos
sys.exit(1)
if cons_input1 == []:
cons_input1 = [0.0 for i in range(2 * win + 1)]
print >> sys.stderr, 'WARNING: PhyloP7 data not found for position', ichr, pos
try:
model = joblib.load(modfile)
except:
print >> sys.stderr, 'ERROR: Program not able to load modfile. Please check that you have installed a compatible version joblib.'
sys.exit(1)
p_cod = 0
cod = 'No'
if r_cod != []:
p_cod = 1
cod = 'Yes'
if pklcod == '':
X = [seq_input + cons_input1 + cons_input2]
y_pred, y_fdrs, c_pred = prediction(X, model)
v_fdr = [y_fdrs[0][0], y_fdrs[0][1]]
else:
X = [seq_input + cons_input1 + cons_input2 + [lwt, lnw, p_cod]]
y_pred, y_fdrs, c_pred = prediction(X, model)
v_fdr = [y_fdrs[0][2], y_fdrs[0][3]]
if y_pred == []:
print >> sys.stderr, 'WARNING: Variants not scored. Check modfile and input'
print '\t'.join([str(i) for i in [ichr, ipos, wt, nw]
]) + '\tNA\tNA\tNA\tNA\tNA\tNA'
else:
print "#CHROM\tPOS\tREF\tALT\tCODING\tPREDICTION\tSCORE\tFDR\tPhyloP100\tAvgPhyloP100"
pp100 = cons_input2[win]
avgpp100 = sum(cons_input2) / float(len(cons_input2))
if c_pred[0] == "Pathogenic": d_fdr = v_fdr[0]
if c_pred[0] == "Benign": d_fdr = v_fdr[1]
print '\t'.join(
str(i) for i in [
ichr, ipos, wt, nw, cod, c_pred[0],
'%.3f' % y_pred[0],
'%.3f' % d_fdr,
'%.3f' % pp100,
'%.3f' % avgpp100
])
return