def outlier():
scores = rutl.removeHeteroChromatin(rutl.loadScores())
field = comale;
df = sort(utl.scanGenome(scores.abs(), {field: lambda x: x.abs().mean(), 'Num. of SNPs': lambda x: x.size}))[
[field, 'Num. of SNPs']]
a = df.iloc[:, 0]
a = a.rename('Global Outliers');
o = a[a > a.quantile(0.99)]
o.to_pickle(utl.outpath + 'real/outliers.global.df')
fig = plt.figure(figsize=(7, 1.5), dpi=300);
pplt.Manhattan(data=a, Outliers=pd.DataFrame(o), fig=fig, markerSize=2, ticksize=8, sortedAlready=True);
[pplt.setSize(ax, 5) for ax in fig.get_axes()];
plt.gcf().subplots_adjust(bottom=0.15);
plt.savefig(utl.paperPath + 'new/{}.pdf'.format('global'))
a = a.rename('Chrom Outliers');
o = a.groupby(level=0).apply(lambda x: x[x > x.quantile(0.99)].loc[x.name])
o.to_pickle(utl.outpath + 'real/outliers.chrom.df')
fig = plt.figure(figsize=(7, 1.5), dpi=300);
pplt.Manhattan(data=a, Outliers=pd.DataFrame(o), fig=fig, markerSize=2, ticksize=8, sortedAlready=True);
[pplt.setSize(ax, 5) for ax in fig.get_axes()]
plt.gcf().subplots_adjust(bottom=0.15);
plt.savefig(utl.paperPath + 'new/{}.pdf'.format('chrom'))
a = a.rename('Local Outliers');
o = localOutliers(a)
o.to_pickle(utl.outpath + 'real/outliers.local.df')
fig = plt.figure(figsize=(7, 1.5), dpi=300);
pplt.Manhattan(data=a, Outliers=pd.DataFrame(o), fig=fig, markerSize=2, ticksize=8, sortedAlready=True);
[pplt.setSize(ax, 5) for ax in fig.get_axes()]
plt.gcf().subplots_adjust(bottom=0.15);
plt.savefig(utl.paperPath + 'new/{}.pdf'.format('local'))
def findCandidateSNPs(outlierMode='local'):
print 'running ', outlierMode
ann = loadANN()["Annotation Annotation_Impact Gene_Name Gene_ID".split()]
intervals = utl.BED.getIntervals(pd.read_pickle(utl.outpath + 'real/outliers.{}.df'.format(outlierMode)), padding=25000)
scores = rutl.removeHeteroChromatin(rutl.loadScores()).rename('H')
scores.shape
candidates = []
for _, row in intervals.iterrows():
row = pd.DataFrame(row).T
row.index.name = 'CHROM'
snp = utl.BED.intersection(scores.reset_index(), row, 'H').rename(columns={'name': 'H', 'start': 'POS'})[
['POS', 'H']].set_index('POS', append=True)['H'].astype(float)
snp = snp[snp > snp.quantile(0.99)]
candidates += [snp]
print snp.shape, row.iloc[0].loc['len']
candidates = pd.DataFrame(pd.concat(candidates)).join(ann, how='inner')
candidates = candidates[candidates['Annotation_Impact'] != 'LOW']
candidates.to_pickle(utl.outpath + 'real/{}.df'.format('cand.' + outlierMode))
def createGwandaDataNew():
def save(df, name='candidatesnps.txt'):
df.sort_index().reset_index().to_csv(utl.outpath + 'real/gowinda/{}.txt'.format(name), sep='\t', header=None,
index=False)
scores = rutl.removeHeteroChromatin(rutl.loadScores()).rename('H')
save(scores, 'allsnps')
for outlierMode in ['local', 'global', 'chrom']:
a = pd.read_pickle(utl.outpath + 'real/{}.df'.format('cand.' + outlierMode))['H'].reset_index().drop_duplicates(
subset=['CHROM', 'POS']).set_index(['CHROM', 'POS']).H
save(a, 'cand.' + outlierMode + '.damped.0')
print a.shape
ann = loadANN()["Annotation Annotation_Impact Gene_Name Gene_ID".split()]
for dampn in [100, 500, 1000, 2000]:
damp = scores.sort_values(ascending=False).iloc[:dampn]
damp = pd.DataFrame(damp).join(ann, how='inner')
damp = \
damp[damp['Annotation_Impact'] != 'LOW']['H'].reset_index().drop_duplicates().set_index(['CHROM', 'POS'])['H']
damp.shape
for outlierMode in ['local', 'global', 'chrom']:
a = pd.read_pickle(utl.outpath + 'real/{}.df'.format('cand.' + outlierMode))[
'H'].reset_index().drop_duplicates(subset=['CHROM', 'POS']).set_index(['CHROM', 'POS']).H
a = pd.concat([a, damp]).reset_index().drop_duplicates(subset=['CHROM', 'POS']).set_index(
['CHROM', 'POS']).H
save(a, 'cand.' + outlierMode + '.damped.{}'.format(dampn))
print a.shape
Genes = pd.read_pickle(utl.outpath + 'real/GO.df').set_index('GO')
Genes = Genes[Genes.AnnID.apply(lambda x: x[:2] == 'CG')]
Genes.AnnID.value_counts()
df = pd.concat([Genes.term.drop_duplicates(), Genes.AnnID.groupby(level=0).apply(lambda x: ' '.join(x.tolist()))],
axis=1)
df.to_csv(utl.outpath + 'real/gowinda/goassociation.CG', sep='\t', header=None)
df = pd.concat([Genes.term.drop_duplicates(), Genes.FBgn.groupby(level=0).apply(lambda x: ' '.join(x.tolist()))],
axis=1)
df.to_csv(utl.outpath + 'real/gowinda/goassociation.FBgn', sep='\t', header=None)
