def read_dist_out(self, SNP_Z):
gene_SNP_out_w = dict()
file = P.sed_file
in_ = open(file)
while True:
ln = in_.readline()
if ln == '':
break
sln1 = ln.split()
ln = in_.readline()
sln2 = ln.split()
if int(sln1[1]) in {1, 2}:
if sln1[3] not in gene_SNP_out_w:
gene_SNP_out_w[sln1[3]] = list()
if sln2[3] not in gene_SNP_out_w:
gene_SNP_out_w[sln2[3]] = list()
if sln1[2] in SNP_Z:
gene_SNP_out_w[sln1[3]].append((sln1[2], U.weight_f(int(sln1[5]))))
if sln2[2] in SNP_Z:
gene_SNP_out_w[sln2[3]].append((sln2[2], U.weight_f(int(sln2[5]))))
if int(sln1[1]) == 3:
if sln1[3] not in gene_SNP_out_w:
gene_SNP_out_w[sln1[3]] = list()
if sln1[2] in SNP_Z:
gene_SNP_out_w[sln1[3]].append((sln1[2], U.weight_f(int(sln1[5]))))
if int(sln2[1]) == 3:
if sln2[3] not in gene_SNP_out_w:
gene_SNP_out_w[sln2[3]] = list()
if sln2[2] in SNP_Z:
gene_SNP_out_w[sln2[3]].append((sln2[2], U.weight_f(int(sln2[5]))))
print('Distances SNP to exons, SNP outside exons and inside genes, from', file)
return gene_SNP_out_w
def create_seq(self):
self.seq_pos = U.read_exon()
self.SNP_pos, _ = U.read_SNP()
for c in range(24):
self.seq_pos[c].extend(self.SNP_pos[c])
[self.seq_pos[i].sort() for i in range(24)]
self.SNP_io = self.SNP_in_gene()
def fit_distance(self):
print('Distance Between SNP, all SNP')
dist = self.read_distance()
cx, cy = U.cumul_number(dist)
param = U.curve_fit(cx, cy, U.gamma_cdf)
U.gamma_draw_fit(cx,
cy,
U.gamma_cdf,
param,
title='Fit distance between SNP, ' +
P.datas[P.data_src])
def fit_lin_log(self, inf_pv, sup_pv):
print(
'\nStep 2- Compute correlation coefficient between log of SNP p-values'
)
print('Fixed limits of p-values', P.inf_pv, P.sup_pv)
self.SNP_pv.sort()
cx, cy = U.cumul_number(self.SNP_pv)
cx, cy = cx[1:], cy[1:]
i_min = np.where(cx > inf_pv)[0][0]
i_max = np.where(cx < sup_pv)[0][-1]
lx = np.log10(cx[i_min:i_max])
ly = np.log10(cy[i_min:i_max])
plt.figure()
plt.title(
P.data +
' linear adjustment of log p-value and log cumulative number')
plt.xticks(ticks=[], labels=[])
plt.yticks(ticks=[], labels=[])
plt.plot(lx, ly)
slope, intercept, r_value, p_value, std_err = stats.linregress(lx, ly)
plt.plot(lx,
lx * slope + intercept,
label='R=' + '{:.4f}'.format(r_value))
plt.legend()
print('number of p-values', len(lx), 'out of a total of',
len(self.SNP_pv))
print('Real limits of p_values: [',
cx[i_min],
', ',
cx[i_max],
']',
sep='')
print('r_value:', r_value)
print('p_value:', p_value)
print('std_err:', std_err)
def outliers(self):
SNP_pv = self.SNP_pv
print(
'Step 1- Choice of fork of p-values to eliminate outlier p-values')
print('Quantiles are computed for p-values in', '[', P.inf_pv, ', ',
P.sup_pv, ']')
print(
'Once choosen, update the parameter to compute correlation coefficient at step 2'
)
print(
'Warning: Python cannot compute the Z-score for p-value < 5.552e-17'
)
quant = [
0.0001, 0.0005, 0.001, 0.005, 0.01, 0.05, 0.1, 0.15, 0.2, 0.8,
0.85, 0.9, 0.95, 0.99, 0.995, 0.999
]
print('quantile\tvalue')
for q in quant:
print(q, np.quantile(SNP_pv, q), sep='\t')
U.plot_histo(SNP_pv, 1000, title=P.data + ' Histogram of p-values')
def SNP_in_gene(self):
seq_pos = U.read_gene()
SNP_io = [dict() for i in range(24)]
for c in range(24):
seq_pos[c].extend(self.SNP_pos[c])
[seq_pos[i].sort() for i in range(24)]
for c in range(24):
in_gene = set()
for sq in seq_pos[c]:
if sq[1] == 1:
in_gene.add(sq[2])
elif sq[1] == 2:
in_gene.remove(sq[2])
else:
SNP_io[c][sq[2]] = set(in_gene)
return SNP_io
def SNP_compute_Z():
# log = True to get details of excluded p-values
log = False
_, SNP_pv = U.read_SNP(log)
out = open(P.pvZ_file, mode='w')
print('Infinity result of computing Z score')
inf_pos, inf_neg = 0, 0
for snp in SNP_pv:
z = stats.norm.ppf(1 - SNP_pv[snp])
if math.isinf(z):
if z == math.inf:
inf_pos += 1
else:
inf_neg += 1
print(snp, z, SNP_pv[snp], sep='\t')
else:
out.write(snp + '\t' + str(SNP_pv[snp]) + '\t' + str(z) + '\n')
print('Z computed from p-values of SNP:', inf_pos, '+inf, ', inf_neg, '-inf /', len(SNP_pv))
out.close()
def __init__(self):
_, SNP_pv_dict = U.read_SNP(log=log)
self.SNP_pv = list(SNP_pv_dict.values())