def read_gm(opts, logger):
opts = copy.deepcopy(opts)
opts.infp = opts.exposure_gm
header = cleangwas.read_header(opts.infp, opts.sep)
cnames = cleangwas.parse_header(header, opts, logger,
cleangwas.default_cnames)
opts.rmpali = False
opts.rmindel = False
opts.unique = False
df = cleangwas.qc(opts, cnames, logger)
return df
def read_exp(opts, logger):
opts = copy.deepcopy(opts)
opts.infp = opts.exposure
header = cleangwas.read_header(opts.infp, opts.sep)
cnames = cleangwas.parse_header(header, opts, logger,
cleangwas.default_cnames)
df = cleangwas.qc(opts, cnames, logger)
df = cleangwas.selectSNP(opts, df, logger)
df = cleangwas.resort_col(df)
df = cleangwas.truncate(opts, df)
return df
def read_outc(opts, logger):
for k, v in cleangwas.del_no.items():
cleangwas.del_no[k] = 0
opts = copy.deepcopy(opts)
opts.infp = opts.outcome
opts.pThresh = 1
opts.include = 'include'
header = cleangwas.read_header(opts.infp, opts.sep)
cnames = cleangwas.parse_header(header, opts, logger,
cleangwas.default_cnames)
insnps = pd.read_csv(opts.include, header=None, names=["RS"], sep='\s+')
insnps['RS'] = insnps['RS'].str.upper()
total_df = pd.DataFrame()
converter = cleangwas.get_converter(cnames)
for chunk in pd.read_csv(opts.infp,
sep=opts.sep,
header=0,
names=cnames,
dtype=str,
iterator=True,
chunksize=2000000):
for k, v in converter.items():
chunk[k] = chunk[k].apply(v)
if 'RS' in chunk.columns:
chunk['RS'] = chunk['RS'].str.upper()
chunk = chunk[chunk['RS'].isin(insnps['RS'])]
if not chunk.empty:
total_df = total_df.append(chunk, sort=False)
else:
logger.logger.warning(
'Can not identify RS column for {0}\nExiting'.format(
opts.infp))
exit(1)
if not total_df.empty:
content = total_df.to_csv(path_or_buf=None,
sep='\t',
na_rep='NA',
float_format='%g',
encoding='utf-8',
index=False)
opts.infp = StringIO(content)
df = cleangwas.qc(opts, cnames, logger)
df = cleangwas.selectSNP(opts, df, logger)
df = cleangwas.resort_col(df)
df = cleangwas.truncate(opts, df)
else:
logger.logger.info('No common SNPs among exposure and outcome.')
df = pd.DataFrame()
try:
os.remove('include')
except:
pass
return df
def read_outc(opts, logger):
for k, v in cleangwas.del_no.items():
cleangwas.del_no[k] = 0
opts = copy.deepcopy(opts)
opts.infp = opts.outcome
opts.pThresh = 1
opts.include = 'include'
header = cleangwas.read_header(opts.infp, opts.sep)
cnames = cleangwas.parse_header(header, opts, logger,
cleangwas.default_cnames)
df = cleangwas.qc(opts, cnames, logger)
df = cleangwas.selectSNP(opts, df, logger)
df = cleangwas.resort_col(df)
df = cleangwas.truncate(opts, df)
try:
os.remove('include')
except:
pass
return df
def mr4gm(opts, logger):
opts = copy.deepcopy(opts)
opts.infp = opts.exposure_gm
header = cleangwas.read_header(opts.infp, opts.sep)
cnames = cleangwas.parse_header(header, opts, logger,
cleangwas.default_cnames)
all_df = pd.read_csv(opts.exposure_gm,
header=0,
names=cnames,
sep='\s+',
dtype=str)
all_df['RS'].to_csv('include',
header=False,
sep='\t',
na_rep='NA',
float_format='%g',
encoding='utf-8',
index=False)
out_df = read_outc(opts, logger)
mr_results = []
mr_heters = []
mr_pleios = []
mr_datas = []
mr_pleiosnps = []
for phe in [x for x in all_df.columns if x in ['Phenotype']]:
#for phe in [x for x in all_df.columns if x in [ 'Phenotype', 'Category']]:
for x in set(all_df.loc[:, phe]):
all_df[all_df[phe] == x].copy().to_csv(
path_or_buf='.exposure_gm.txt',
sep='\t',
na_rep='NA',
float_format='%g',
encoding='utf-8',
index=False)
out_df.to_csv(path_or_buf='.outcome_gm.txt',
sep='\t',
na_rep='NA',
float_format='%g',
encoding='utf-8',
index=False)
opts.exposure = '.exposure_gm.txt'
opts.outcome = '.outcome_gm.txt'
opts.exp_name = x
opts.exposure_gm = None
mr_result, mr_heter, mr_pleio, mr_data, mr_pleiosnp = mr4general(
opts, logger)
mr_results.append(mr_result)
mr_heters.append(mr_heter)
mr_pleios.append(mr_pleio)
mr_datas.append(mr_data)
mr_pleiosnps.append(mr_pleiosnp)
try:
os.remove('.exposure_gm.txt')
os.remove('.outcome_gm.txt')
except:
pass
if len([x for x in mr_results if not x.empty]) != 0:
mr_results = pd.concat([x for x in mr_results if not x.empty],
axis=0,
ignore_index=True,
sort=False,
copy=True)
if len([x for x in mr_heters if not x.empty]) != 0:
mr_heters = pd.concat([x for x in mr_heters if not x.empty],
axis=0,
ignore_index=True,
sort=False,
copy=True)
if len([x for x in mr_pleios if not x.empty]) != 0:
mr_pleios = pd.concat([x for x in mr_pleios if not x.empty],
axis=0,
ignore_index=True,
sort=False,
copy=True)
if len([x for x in mr_datas if not x.empty]) != 0:
mr_datas = pd.concat([x for x in mr_datas if not x.empty],
axis=0,
ignore_index=True,
sort=False,
copy=True)
if opts.heterogeneity != 'none' and len(
[x for x in mr_pleiosnps if not x.empty]) != 0:
mr_pleiosnps = pd.concat([x for x in mr_pleiosnps if not x.empty],
axis=0,
ignore_index=True,
sort=False,
copy=True)
else:
mr_pleiosnps = pd.DataFrame()
return mr_results, mr_heters, mr_pleios, mr_datas, mr_pleiosnps