def main(args):
logging.basicConfig(level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log")])
logging.info(args)
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.create_dir_if_not_exist(join(args.out_expr_dir,'raw'))
utils.create_dir_if_not_exist(join(args.out_expr_dir,'processed'))
gse = GEOparse.get_GEO(geo='GSE16032', destdir=join(args.out_expr_dir,'raw'))
annotated = gse.pivot_and_annotate('VALUE', gse.gpls['GPL570'], 'ENTREZ_GENE_ID')
annotated2 = annotated[~pd.isnull(annotated.ENTREZ_GENE_ID)]
annotated2 = annotated2.loc[~annotated2.isnull().values.all(axis=1)]
annotated2['ENTREZ_GENE_ID'] = annotated2.ENTREZ_GENE_ID.str.split('///').str[0].astype(int)
annotated2 = annotated2.set_index('ENTREZ_GENE_ID')
annotated2 = np.log(annotated2)
disease_cls = ['disease state: Acute', 'disease state: Convalescence']
logging.info(disease_cls)
disease_gsm = [gsm for gsm in gse.gsms if gse.gsms[gsm].metadata['characteristics_ch1'][1] in disease_cls]
logging.info("Disease GSM: {}".format(len(disease_gsm)))
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.write_expr(join(args.out_expr_dir, 'processed', 'expr.tsv'), annotated2)
utils.write_text(join(args.out_expr_dir,'processed','disease_gsms.txt'), disease_gsm)
def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log")])
logging.info(args)
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'raw'))
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'processed'))
gse = GEOparse.get_GEO(geo='GSE64913',
destdir=join(args.out_expr_dir, 'raw'))
annotated = gse.pivot_and_annotate('VALUE', gse.gpls['GPL570'],
'ENTREZ_GENE_ID')
annotated2 = annotated[~pd.isnull(annotated.ENTREZ_GENE_ID)]
annotated2 = annotated2.loc[~annotated2.isnull().values.all(axis=1)]
annotated2['ENTREZ_GENE_ID'] = annotated2.ENTREZ_GENE_ID.str.split(
'///').str[0].astype(int)
annotated2 = annotated2.set_index('ENTREZ_GENE_ID')
classes = {}
classes['healthy_cae'] = [
'diagnosis: Healthy', 'cell type: Central airway epithelium'
]
classes['healthy_pae'] = [
'diagnosis: Healthy', 'cell type: Peripheral airway epithelium'
]
classes['asthma_cae'] = [
'diagnosis: Severe Asthmatic', 'cell type: Central airway epithelium'
]
classes['asthma_pae'] = [
'diagnosis: Severe Asthmatic',
'cell type: Peripheral airway epithelium'
]
logging.info(classes)
gsms = {
cls: [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['characteristics_ch1'][1] == classes[cls]
[0] and gse.gsms[gsm].metadata['characteristics_ch1'][5] ==
classes[cls][1]
]
for cls in classes
}
logging.info(' '.join(
['{} GSM:{}'.format(cls, len(gsms[cls])) for cls in classes]))
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.write_expr(join(args.out_expr_dir, 'processed', 'expr.tsv'),
annotated2)
for cls in classes:
utils.write_text(
join(args.out_expr_dir, 'processed', '{}_gsms.txt'.format(cls)),
gsms[cls])
def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[
logging.FileHandler("../logs/report.log"),
logging.StreamHandler()
])
logging.info(args)
gda = pd.read_csv(args.in_gda_path, sep='\t')
disease_maps = pd.read_csv(args.in_dismap_path, sep='|')
net = utils.read_network(args.in_net_path)
gda = gda[gda.geneId.isin(net.nodes())]
#gda = gda.groupby('diseaseId').filter(lambda x: len(x) >= args.min_disease_size)
cols = [
'diseaseId', 'diseaseName', 'diseaseType', 'diseaseClass',
'diseaseSemanticType', 'geneId'
]
diseases = gda.groupby('diseaseId').apply(lambda x: pd.Series(
{
'diseaseName': x.diseaseName.tolist()[0],
'diseaseType': x.diseaseType.tolist()[0],
'diseaseClass': x.diseaseClass.tolist()[0],
'diseaseSemanticType': x.diseaseSemanticType.tolist()[0],
'n_genes': x.shape[0]
})).reset_index()
print(diseases.shape)
diseases = diseases[(diseases.n_genes > args.min_disease_size)
& (diseases.n_genes < args.max_disease_size) &
(diseases.diseaseType == 'disease') &
(diseases.diseaseSemanticType
== 'Disease or Syndrome')]
print(diseases.shape)
def get_doids(diseaseId):
return '|'.join([
'DOID:{}'.format(doid)
for doid in disease_maps[(disease_maps.vocabulary == 'DO') & (
disease_maps.diseaseId == diseaseId)].code.tolist()
])
diseases['DOID'] = diseases.diseaseId.apply(get_doids)
doids = list(
set([
'DOID:{}'.format(doid)
for doid in disease_maps[(disease_maps.vocabulary == 'DO')
& (disease_maps.diseaseId.isin(
diseases.diseaseId))].code.tolist()
]))
gda.to_csv(args.out_gda_file, sep='\t', index=False)
diseases.to_csv(args.out_disease_file, sep='\t', index=False)
utils.write_text(args.out_doid_file, doids)
def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log")])
logging.info(args)
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'raw'))
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'processed'))
gse = GEOparse.get_GEO(geo='GSE54837',
destdir=join(args.out_expr_dir, 'raw'))
annotated = gse.pivot_and_annotate('VALUE', gse.gpls['GPL570'],
'ENTREZ_GENE_ID')
annotated2 = annotated[~pd.isnull(annotated.ENTREZ_GENE_ID)]
annotated2 = annotated2.loc[~annotated2.isnull().values.all(axis=1)]
annotated2['ENTREZ_GENE_ID'] = annotated2.ENTREZ_GENE_ID.str.split(
'///').str[0].astype(int)
annotated2 = annotated2.set_index('ENTREZ_GENE_ID')
annotated2[annotated2 <= 0] = 0.001
annotated2 = np.log(annotated2)
disease_cls = ['subject type: COPD Subjects']
healthy_cls = [
'subject type: Non-smoker Controls', 'subject type: Smoker Controls'
]
healthy_non_smoker_cls = ['subject type: Non-smoker Controls']
logging.info(disease_cls)
logging.info(healthy_cls)
logging.info(healthy_non_smoker_cls)
disease_gsm = [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['characteristics_ch1'][5] in disease_cls
]
healthy_gsm = [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['characteristics_ch1'][5] in healthy_cls
]
healthy_non_smoker_gsm = [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['characteristics_ch1'][5] in
healthy_non_smoker_cls
]
logging.info(
"Disease GSM: {}, Healthy GSM: {}, Healthy non smoker GSM: {}".format(
len(disease_gsm), len(healthy_gsm), len(healthy_non_smoker_gsm)))
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.write_expr(join(args.out_expr_dir, 'processed', 'expr.tsv'),
annotated2)
utils.write_text(join(args.out_expr_dir, 'processed', 'disease_gsms.txt'),
disease_gsm)
utils.write_text(join(args.out_expr_dir, 'processed', 'healthy_gsms.txt'),
healthy_gsm)
def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log")])
logging.info(args)
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'raw'))
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'processed'))
gse = GEOparse.get_GEO(geo='GSE473',
destdir=join(args.out_expr_dir, 'raw'))
annotated = gse.pivot_and_annotate('VALUE', gse.gpls['GPL96'],
'ENTREZ_GENE_ID')
gsm96 = [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['platform_id'][0] == 'GPL96'
] # we choose only GPL96 platform
annotated2 = annotated[gsm96 + ['ENTREZ_GENE_ID']]
print(annotated2.shape)
annotated2 = annotated2[~pd.isnull(annotated2.ENTREZ_GENE_ID)]
print(annotated2.shape)
annotated2 = annotated2.loc[~annotated2.isnull().values.any(axis=1)]
print(annotated2.shape)
annotated2['ENTREZ_GENE_ID'] = annotated2.ENTREZ_GENE_ID.str.split(
'///').str[0].astype(int)
annotated2 = annotated2.set_index('ENTREZ_GENE_ID')
annotated2 = np.log(annotated2)
#{'astM_atop', 'astM_nonatop', 'astS_atop', 'ctr_atop', 'ctr_nonatop'}
classes = {}
classes['asthma_med_nonatop'] = ['astM_nonatop']
classes['control_nonatop'] = ['ctr_nonatop']
logging.info(classes)
gsms = {
cls: [
gsm for gsm in gsm96
if classes[cls][0] in gse.gsms[gsm].metadata['title'][0]
]
for cls in classes
}
logging.info(' '.join(
['{} GSM:{}'.format(cls, len(gsms[cls])) for cls in classes]))
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.write_expr(join(args.out_expr_dir, 'processed', 'expr.tsv'),
annotated2)
for cls in classes:
utils.write_text(
join(args.out_expr_dir, 'processed', '{}_gsms.txt'.format(cls)),
gsms[cls])
def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log")])
logging.info(args)
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'raw'))
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'processed'))
gse = GEOparse.get_GEO(geo='GSE31773',
destdir=join(args.out_expr_dir, 'raw'))
annotated = gse.pivot_and_annotate('VALUE', gse.gpls['GPL570'],
'ENTREZ_GENE_ID')
annotated2 = annotated[~pd.isnull(annotated.ENTREZ_GENE_ID)]
annotated2 = annotated2.loc[~annotated2.isnull().values.all(axis=1)]
annotated2['ENTREZ_GENE_ID'] = annotated2.ENTREZ_GENE_ID.str.split(
'///').str[0].astype(int)
annotated2 = annotated2.set_index('ENTREZ_GENE_ID')
classes = {}
classes['cd4_severe'] = ['CD4_Tcells_severe_asthma']
classes['cd8_severe'] = ['CD8_Tcells_severe_asthma']
classes['cd4_healthy'] = ['CD4_Tcells_healthy_donor']
classes['cd8_healthy'] = ['CD8_Tcells_healthy_donor']
classes['cd8_non_severe'] = ['CD8_Tcells_non_severe_asthma']
classes['cd4_non_severe'] = ['CD4_Tcells_non_severe_asthma']
classes['asthma_severe'] = [
'CD4_Tcells_severe_asthma', 'CD8_Tcells_severe_asthma'
]
classes['healthy'] = [
'CD4_Tcells_healthy_donor', 'CD8_Tcells_healthy_donor'
]
logging.info(classes)
gsms = {
cls: [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['source_name_ch1'][0] in classes[cls]
]
for cls in classes
}
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.write_expr(join(args.out_expr_dir, 'processed', 'expr.tsv'),
annotated2)
for cls in classes:
logging.info("{} GSM: {}".format(cls, len(gsms[cls])))
utils.write_text(
join(args.out_expr_dir, 'processed', '{}_gsms.txt'.format(cls)),
gsms[cls])
def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log")])
logging.info(args)
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'raw'))
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'processed'))
gse = GEOparse.get_GEO(geo='GSE16972',
destdir=join(args.out_expr_dir, 'raw'))
annotated = gse.pivot_and_annotate('VALUE', gse.gpls['GPL96'],
'ENTREZ_GENE_ID')
print(annotated.shape)
annotated2 = annotated[~pd.isnull(annotated.ENTREZ_GENE_ID)]
print(annotated2.shape)
annotated2 = annotated2.loc[~annotated2.isnull().values.any(axis=1)]
print(annotated2.shape)
annotated2['ENTREZ_GENE_ID'] = annotated2.ENTREZ_GENE_ID.str.split(
'///').str[0].astype(int)
annotated2 = annotated2.set_index('ENTREZ_GENE_ID')
annotated2 = np.log(annotated2)
classes = {}
classes['copd'] = ['disease status: COPD patient']
classes['control'] = ['disease status: control patient']
logging.info(classes)
gsms = {
cls: [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['characteristics_ch1'][0] ==
'cell type: alveolar macrophage' and
gse.gsms[gsm].metadata['characteristics_ch1'][1] == classes[cls][0]
]
for cls in classes
}
logging.info(' '.join(
['{} GSM:{}'.format(cls, len(gsms[cls])) for cls in classes]))
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.write_expr(join(args.out_expr_dir, 'processed', 'expr.tsv'),
annotated2)
for cls in classes:
utils.write_text(
join(args.out_expr_dir, 'processed', '{}_gsms.txt'.format(cls)),
gsms[cls])
def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log")])
logging.info(args)
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'raw'))
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'processed'))
gse = GEOparse.get_GEO(geo='GSE18965',
destdir=join(args.out_expr_dir, 'raw'))
annotated = gse.pivot_and_annotate('VALUE', gse.gpls['GPL96'],
'ENTREZ_GENE_ID')
annotated2 = annotated[~pd.isnull(annotated.ENTREZ_GENE_ID)]
annotated2 = annotated2.loc[~annotated2.isnull().values.all(axis=1)]
annotated2['ENTREZ_GENE_ID'] = annotated2.ENTREZ_GENE_ID.str.split(
'///').str[0].astype(int)
annotated2 = annotated2.set_index('ENTREZ_GENE_ID')
disease_cls = [
gse.gsms[gsm].metadata['title'][0] for gsm in gse.gsms
if 'AA' in gse.gsms[gsm].metadata['title'][0]
]
healthy_cls = [
gse.gsms[gsm].metadata['title'][0] for gsm in gse.gsms
if 'HN' in gse.gsms[gsm].metadata['title'][0]
]
logging.info(disease_cls)
logging.info(healthy_cls)
disease_gsm = [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['title'][0] in disease_cls
]
healthy_gsm = [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['title'][0] in healthy_cls
]
logging.info("Disease GSM: {}, Healthy GSM: {}".format(
len(disease_gsm), len(healthy_gsm)))
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.write_expr(join(args.out_expr_dir, 'processed', 'expr.tsv'),
annotated2)
utils.write_text(join(args.out_expr_dir, 'processed', 'disease_gsms.txt'),
disease_gsm)
utils.write_text(join(args.out_expr_dir, 'processed', 'healthy_gsms.txt'),
healthy_gsm)
def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log")])
logging.info(args)
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'raw'))
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'processed'))
gse = GEOparse.get_GEO(geo='GSE89809',
destdir=join(args.out_expr_dir, 'raw'))
annotated = gse.pivot_and_annotate('VALUE', gse.gpls['GPL13158'],
'ENTREZ_GENE_ID')
annotated2 = annotated[~pd.isnull(annotated.ENTREZ_GENE_ID)]
annotated2 = annotated2.loc[~annotated2.isnull().values.all(axis=1)]
annotated2['ENTREZ_GENE_ID'] = annotated2.ENTREZ_GENE_ID.str.split(
'///').str[0].astype(int)
annotated2 = annotated2.set_index('ENTREZ_GENE_ID')
# ['Severe_Spm', 'Healthy_Spm', 'Healthy_Epithelial', 'Mild_Epithelial', 'Severe_Epithelial', 'Mild_Spm', 'Healthy_BAL',
# 'Moderate_Epithelial', 'Severe_BAL', 'Mild_BAL', 'Moderate_Spm', 'Moderate_BAL']
classes = set([
'_'.join(gse.gsms[gsm].metadata['title'][0].split('_')[2:4])
for gsm in gse.gsms
])
logging.info(classes)
gsms = {
cls:
[gsm for gsm in gse.gsms if cls in gse.gsms[gsm].metadata['title'][0]]
for cls in classes
}
logging.info(' '.join(
['{} GSM:{}'.format(cls, len(gsms[cls])) for cls in classes]))
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.write_expr(join(args.out_expr_dir, 'processed', 'expr.tsv'),
annotated2)
for cls in classes:
utils.write_text(
join(args.out_expr_dir, 'processed', '{}_gsms.txt'.format(cls)),
gsms[cls])
def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log")])
logging.info(args)
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'raw'))
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'processed'))
gse = GEOparse.get_GEO(geo='GSE37147',
destdir=join(args.out_expr_dir, 'raw'))
annotated = gse.pivot_and_annotate('VALUE', gse.gpls['GPL13243'],
'SPOT_ID')
annotated2 = annotated.rename(columns={'SPOT_ID': 'ENTREZ_GENE_ID'})
annotated2 = annotated2.set_index('ENTREZ_GENE_ID')
disease_cls = ['copd: yes']
healthy_cls = ['copd: no']
logging.info(disease_cls)
logging.info(healthy_cls)
disease_gsm = [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['characteristics_ch1'][4] in disease_cls
]
healthy_gsm = [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['characteristics_ch1'][4] in healthy_cls
]
disease_no_asthma_gsm = [gsm for gsm in gse.gsms \
if gse.gsms[gsm].metadata['characteristics_ch1'][4] in disease_cls \
and gse.gsms[gsm].metadata['characteristics_ch1'][9] == 'history of asthma: no']
healthy_no_asthma_gsm = [gsm for gsm in gse.gsms \
if gse.gsms[gsm].metadata['characteristics_ch1'][4] in healthy_cls \
and gse.gsms[gsm].metadata['characteristics_ch1'][9] == 'history of asthma: no']
logging.info("Disease GSM: {}, Healthy GSM: {}".format(
len(disease_gsm), len(healthy_gsm)))
logging.info(
"Disease (no history of asthma) GSM: {} Healthy (no history of asthma) GSM: {}"
.format(len(disease_no_asthma_gsm), len(healthy_no_asthma_gsm)))
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.write_expr(join(args.out_expr_dir, 'processed', 'expr.tsv'),
annotated2)
utils.write_text(join(args.out_expr_dir, 'processed', 'disease_gsms.txt'),
disease_gsm)
utils.write_text(join(args.out_expr_dir, 'processed', 'healthy_gsms.txt'),
healthy_gsm)
utils.write_text(
join(args.out_expr_dir, 'processed',
'disease_no_history_asthma_gsms.txt'), disease_no_asthma_gsm)
utils.write_text(
join(args.out_expr_dir, 'processed',
'healthy_no_history_asthma_gsms.txt'), healthy_no_asthma_gsm)
def main(args):
logging.basicConfig(level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log")])
logging.info(args)
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.create_dir_if_not_exist(join(args.out_expr_dir,'raw'))
utils.create_dir_if_not_exist(join(args.out_expr_dir,'processed'))
gse = GEOparse.get_GEO(geo='GSE57148', destdir=join(args.out_expr_dir,'raw'))
with closing(request.urlopen(
'ftp://ftp.ncbi.nlm.nih.gov/geo/series/GSE57nnn/GSE57148/suppl/GSE57148_COPD_FPKM_Normalized.txt.gz')) as r:
with open(join(args.out_expr_dir,'raw','GSE57148_COPD_FPKM_Normalized.txt.gz'), 'wb') as f:
shutil.copyfileobj(r, f)
counts = pd.read_csv(join(args.out_expr_dir,'raw','GSE57148_COPD_FPKM_Normalized.txt.gz'),sep='\t')
annotated2 = counts.rename(columns={gse.gsms[gsm].metadata['title'][0].split(' ')[-1]: gsm for gsm in gse.gsms})
print(annotated2.isnull().values.any(axis=1).sum())
print(annotated2.isnull().values.all(axis=1).sum())
convmatr = utils.gm._sources['biomart'].set_index('symbol')
annotated2['ENTREZ_GENE_ID'] = convmatr.loc[annotated2.GeneName.tolist()].entrez.tolist()
print(annotated2.shape)
annotated2 = annotated2[~annotated2.ENTREZ_GENE_ID.isnull()]
print(annotated2.shape)
annotated2['ENTREZ_GENE_ID'] = annotated2['ENTREZ_GENE_ID'].astype(int)
del annotated2['GeneName']
annotated2 = annotated2.set_index('ENTREZ_GENE_ID')
disease_cls = ['disease state: COPD']
healthy_cls = ['disease state: Normal']
logging.info(disease_cls)
logging.info(healthy_cls)
disease_gsm = [gsm for gsm in gse.gsms if gse.gsms[gsm].metadata['characteristics_ch1'][0] in disease_cls]
healthy_gsm = [gsm for gsm in gse.gsms if gse.gsms[gsm].metadata['characteristics_ch1'][0] in healthy_cls]
logging.info("Disease GSM: {}, Healthy GSM: {}".format(len(disease_gsm), len(healthy_gsm)))
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.write_expr(join(args.out_expr_dir, 'processed', 'expr.tsv'), annotated2)
utils.write_text(join(args.out_expr_dir, 'processed', 'disease_gsms.txt'), disease_gsm)
utils.write_text(join(args.out_expr_dir, 'processed', 'healthy_gsms.txt'), healthy_gsm)
def main(args):
logging.basicConfig(
level=logging.INFO,
format='%(module)s:%(levelname)s:%(asctime)s:%(message)s',
handlers=[logging.FileHandler("../logs/report.log")])
logging.info(args)
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'raw'))
utils.create_dir_if_not_exist(join(args.out_expr_dir, 'processed'))
gse = GEOparse.get_GEO(geo='GSE104468',
destdir=join(args.out_expr_dir, 'raw'))
annotated = gse.pivot_and_annotate('VALUE', gse.gpls['GPL21185'],
'GENE_SYMBOL')
convmatr = utils.gm._sources['biomart'].set_index('symbol')
annotated2 = annotated[~pd.isnull(annotated.GENE_SYMBOL)]
annotated2 = annotated2.loc[~annotated2.isnull().values.all(axis=1)]
annotated2['ENTREZ_GENE_ID'] = convmatr.loc[
annotated2.GENE_SYMBOL.tolist()].entrez.tolist()
annotated2 = annotated2[~annotated2.ENTREZ_GENE_ID.isnull()]
annotated2['ENTREZ_GENE_ID'] = annotated2['ENTREZ_GENE_ID'].astype(int)
del annotated2['GENE_SYMBOL']
annotated2 = annotated2.set_index('ENTREZ_GENE_ID')
classes = {}
classes['asthma_pbmc'] = ['disease state: Asthma', 'cell type: PBMC']
classes['asthma_bronch'] = [
'disease state: Asthma', 'cell type: bronchial epithelia'
]
classes['asthma_nasal'] = [
'disease state: Asthma', 'cell type: nasal epithelia'
]
classes['normal_pbmc'] = ['disease state: Normal', 'cell type: PBMC']
classes['normal_bronch'] = [
'disease state: Normal', 'cell type: bronchial epithelia'
]
classes['normal_nasal'] = [
'disease state: Normal', 'cell type: nasal epithelia'
]
logging.info(classes)
gsms = {
cls: [
gsm for gsm in gse.gsms
if gse.gsms[gsm].metadata['characteristics_ch1'][2] == classes[cls]
[0] and gse.gsms[gsm].metadata['characteristics_ch1'][1] ==
classes[cls][1]
]
for cls in classes
}
logging.info(' '.join(
['{} GSM:{}'.format(cls, len(gsms[cls])) for cls in classes]))
utils.create_dir_if_not_exist(args.out_expr_dir)
utils.write_expr(join(args.out_expr_dir, 'processed', 'expr.tsv'),
annotated2)
for cls in classes:
utils.write_text(
join(args.out_expr_dir, 'processed', '{}_gsms.txt'.format(cls)),
gsms[cls])