for samplename, sample in samples.iterrows():
if VERBOSE >= 1:
print samplename
sample = SamplePat(sample)
pname = sample.patient
ref = sample.get_reference("genomewide", "gb")
# Collect the insertions (where possible)
ics = {}
for fragment in ["F" + str(i) for i in xrange(1, 7)]:
try:
ic = sample.get_insertions(fragment, merge_read_types=False)
except IOError:
continue
start = find_annotation(ref, fragment).location.nofuzzy_start
ics[(fragment, start)] = ic
if not len(ics):
if VERBOSE >= 1:
print "No data found: skipping"
continue
# Merge insertions
ic = merge_insertions(ics, VERBOSE=VERBOSE)
if save_to_file:
fn_out = sample.get_insertions_filename("genomewide")
save_insertions(fn_out, ic)
if VERBOSE >= 1:
print "Genomewide insertions saved to:", fn_out
parser.add_argument('--reference', default='HXB2',
help='Reference to use for alignment')
parser.add_argument('--verbose', type=int, default=0,
help='Verbosity level [0-4]')
parser.add_argument('--subtypes', nargs='+', default=['B'],
help='Subtypes to keep')
args = parser.parse_args()
regions = args.regions
refname = args.reference
VERBOSE = args.verbose
subtypes = args.subtypes
from hivwholeseq.reference import load_custom_reference
from hivwholeseq.utils.sequence import find_annotation
ref = load_custom_reference('HXB2', 'gb')
for region in regions:
regm = np.array(find_annotation(ref, region).extract(ref), 'S1')
for subtype in subtypes:
fn = get_subtype_reference_alignment_filename(region,
subtype=subtype,
refname=refname,
VERBOSE=VERBOSE)
alim = np.array(AlignIO.read(fn, 'fasta'), 'S1')
weird = ((alim != regm).mean(axis=1) > 0.2)
print region, subtype, weird.sum()
help='Reference to use for alignment')
parser.add_argument('--verbose',
type=int,
default=0,
help='Verbosity level [0-4]')
parser.add_argument('--subtypes',
nargs='+',
default=['B'],
help='Subtypes to keep')
args = parser.parse_args()
regions = args.regions
refname = args.reference
VERBOSE = args.verbose
subtypes = args.subtypes
from hivwholeseq.reference import load_custom_reference
from hivwholeseq.utils.sequence import find_annotation
ref = load_custom_reference('HXB2', 'gb')
for region in regions:
regm = np.array(find_annotation(ref, region).extract(ref), 'S1')
for subtype in subtypes:
fn = get_subtype_reference_alignment_filename(region,
subtype=subtype,
refname=refname,
VERBOSE=VERBOSE)
alim = np.array(AlignIO.read(fn, 'fasta'), 'S1')
weird = ((alim != regm).mean(axis=1) > 0.2)
print region, subtype, weird.sum()
def correlate_epitope_substitution(ds, dctl):
'''Correlate presence of a substitution with epitope'''
from hivwholeseq.data.primers import primers_coordinates_HXB2_outer
start_F1 = primers_coordinates_HXB2_outer['F1'][0][1]
end_F6 = primers_coordinates_HXB2_outer['F6'][1][0]
ds = ds.copy()
dg = []
for pcode, datum in dctl.groupby('pcode'):
a = np.arange(start_F1, end_F6)
b = np.zeros(len(a), bool)
for _, epi in datum.iterrows():
b[(a >= epi['start_HXB2']) & (a < epi['end_HXB2'])] = True
c = np.zeros(len(a), bool)
datum = ds.loc[ds['pcode'] == pcode]
# Keep only nonsyn substitutions
datum = datum.loc[datum['syn'] == False]
c[datum['pos_ref'] - a[0]] = True
dat = {
'pos': a,
'epitope': b,
'substitution': c,
}
dat = pd.DataFrame(dat)
dat['pcode'] = pcode
dg.append(dat)
dg = pd.concat(dg)
# Exclude env because it has antibody-related substitutions
from hivwholeseq.reference import load_custom_reference
from hivwholeseq.utils.sequence import find_annotation
ref = load_custom_reference('HXB2', 'gb')
start_env = find_annotation(ref, 'gp41').location.nofuzzy_start
end_env = find_annotation(ref, 'gp41').location.nofuzzy_end - 450
dg = dg.loc[(dg['pos'] < start_env) | (dg['pos'] >= end_env)]
M = dg.groupby(['epitope', 'substitution']).size().unstack()
Ma = np.array(M)
xp = 1.0 * Ma[1, 0] / Ma[0, 0] * Ma[0, 1]
xs = Ma[1, 1] - xp
print M
from scipy.stats import fisher_exact
print 'Fisher\'s exact enrichment:', fisher_exact(Ma)[0]
print 'Fisher\'s exact P value:', fisher_exact(Ma)[1]
print 'expected:', xp
print 'excess:', xs, 'per patient:', xs / 9.0
pos_epi = dg.loc[dg['epitope'] == True]['pos'].unique()
dg2 = dg.loc[dg['pos'].isin(pos_epi)].copy()
M2 = dg2.groupby(['epitope', 'substitution']).size().unstack()
M2a = np.array(M2)
xp = 1.0 * M2a[1, 0] / M2a[0, 0] * M2a[0, 1]
xs = M2a[1, 1] - xp
print M2
print '\nFisher\'s exact enrichment:', fisher_exact(M2a)[0]
print 'Fisher\'s exact P value:', fisher_exact(M2a)[1]
print 'expected:', xp
print 'excess:', xs, 'per patient:', xs / 9.0
return {
'dg': dg,
'dg2': dg2,
}
for samplename, sample in samples.iterrows():
if VERBOSE >= 1:
print samplename
sample = SamplePat(sample)
pname = sample.patient
ref = sample.get_reference('genomewide', 'gb')
# Collect the insertions (where possible)
ics = {}
for fragment in ['F' + str(i) for i in xrange(1, 7)]:
try:
ic = sample.get_insertions(fragment, merge_read_types=False)
except IOError:
continue
start = find_annotation(ref, fragment).location.nofuzzy_start
ics[(fragment, start)] = ic
if not len(ics):
if VERBOSE >= 1:
print 'No data found: skipping'
continue
# Merge insertions
ic = merge_insertions(ics, VERBOSE=VERBOSE)
if save_to_file:
fn_out = sample.get_insertions_filename('genomewide')
save_insertions(fn_out, ic)
if VERBOSE >= 1:
print 'Genomewide insertions saved to:', fn_out
