import sys
import os
from hivwholeseq.utils.generic import mkdirs
from hivwholeseq.patients.samples import itersample
from hivwholeseq.sequencing.samples import load_samples_sequenced as lss
from hivwholeseq.patients.samples import load_samples_sequenced as lssp
from hivwholeseq.sequencing.filenames import get_sample_foldername
# Script
if __name__ == '__main__':
samples_pat = lssp()
samples_seq = lss()
for samplename, sample in itersample(samples_pat):
root_foldername = sample.get_foldername()+'samples_sequencing/'
mkdirs(root_foldername)
for samplenameseq, sampleseq in samples_seq.iterrows():
if sampleseq['patient sample'] == samplename:
src_folder = get_sample_foldername(samplenameseq)
dst_folder = root_foldername+samplenameseq
if not os.path.islink(dst_folder):
os.symlink(src_folder, dst_folder)
print 'Symlink:', src_folder, dst_folder
else:
print 'Esists:', dst_folder
samples = lssp()
if pnames is not None:
samples = samples.loc[samples.patient.isin(pnames)]
elif samplenames is not None:
samples = samples.loc[samples.index.isin(samplenames)]
if VERBOSE >= 2:
print "samples", samples.index.tolist()
if use_plot:
fig, ax = plt.subplots()
for ir, region in enumerate(regions):
for isam, (samplename, sample) in enumerate(itersample(samples)):
if VERBOSE >= 1:
print samplename, region
cov = sample.get_coverage(region)
if use_plot:
color = cm.jet(1.0 * (isam + ir * len(samples)) / (len(regions) * len(samples)))
ax.plot(cov + 0.1, lw=2, c=color, label=samplename + ", " + region)
if use_plot:
ax.set_xlabel("Position [bp]")
ax.set_ylabel("Coverage")
ax.set_yscale("log")
ax.set_ylim(0.1, 1e6)
samples = lssp()
if pnames is not None:
samples = samples.loc[samples.patient.isin(pnames)]
elif samplenames is not None:
samples = samples.loc[samples.index.isin(samplenames)]
if VERBOSE >= 2:
print 'samples', samples.index.tolist()
if use_plot:
fig, ax = plt.subplots()
for ir, region in enumerate(regions):
for isam, (samplename, sample) in enumerate(itersample(samples)):
if VERBOSE >= 1:
print samplename, region
cov = sample.get_coverage(region)
if use_plot:
color = cm.jet(1.0 * (isam + ir * len(samples)) /
(len(regions) * len(samples)))
ax.plot(cov + 0.1,
lw=2,
c=color,
label=samplename + ', ' + region)
if use_plot:
ax.set_xlabel('Position [bp]')
author: Fabio Zanini
date: 18/05/15
content: Get days since infection from dates.
'''
# Globals
import pandas as pd
from hivwholeseq.patients.patients import load_patients
from hivwholeseq.patients.samples import load_samples_sequenced, itersample
# Script
if __name__ == '__main__':
samples = load_samples_sequenced(include_empty=True)
patients = load_patients()
dsi = []
for samplename, sample in itersample(samples):
t = sample.date
t0 = patients.loc[sample.patient]['infect date best']
dt = (t - t0).days
dsi.append({'sample': samplename,
'dsi': dt,
})
dsi = pd.DataFrame(dsi).set_index('sample')
print dsi
