def load_rnaseq_data(disease_code, with_clinical=False, wait_time=30):
# Fetch RNA data
archive_path = prefetch_rnaseq_data(disease_code)
result_dir = os.path.join(os.path.dirname(archive_path), disease_code,
'gene_expression')
if not os.path.exists(result_dir):
os.makedirs(result_dir)
# Unpack tar file
archive = tarfile.open(archive_path)
gene_quantification_files = set(
el for el in archive if 'genes.normalized_results' in el.name)
if len(gene_quantification_files.difference(
os.listdir(result_dir))) > 0:
archive.extractall(members=gene_quantification_files,
path=result_dir)
archive.extract('FILE_SAMPLE_MAP.txt', path=result_dir)
# Load map from samples to RNA files
rna_file_sample_map = pd.read_csv(os.path.join(result_dir,
'FILE_SAMPLE_MAP.txt'),
sep='\t')
rna_file_sample_map_id_split = rna_file_sample_map[
'barcode(s)'].str.rsplit('-', n=4, expand=True)
rna_file_sample_map_id_split.columns = [
'TCGA_ID', 'SampleID', 'PortionID', 'PlateID', 'CenterID'
]
rna_file_sample_map = rna_file_sample_map.join(
rna_file_sample_map_id_split)
gene_filter = rna_file_sample_map['filename'].str.contains(
'genes.normalized_results')
gene_rna_file_sample_map = rna_file_sample_map[gene_filter]
rna_dfs = []
for (f, sample) in list(
zip(gene_rna_file_sample_map['filename'],
gene_rna_file_sample_map['TCGA_ID'])):
sample_rna_df = pd.read_csv(os.path.join(result_dir, f), sep='\t')
sample_rna_df['TCGA_ID'] = sample
sample_rna_df['gene_name'] = sample_rna_df.gene_id.str.split(
'|').str.get(0)
rna_dfs.append(sample_rna_df)
rna_df = pd.concat(rna_dfs, copy=False).merge(rna_file_sample_map)
if with_clinical:
patient_data_df = load_clinical_data(disease_code)
merged = rna_df.merge(patient_data_df,
how='outer',
left_on='TCGA_ID',
right_on='bcr_patient_barcode')
return merged
else:
return rna_df
def load_rnaseq_data(disease_code,
with_clinical=False,
wait_time=30):
# Fetch RNA data
archive_path = prefetch_rnaseq_data(disease_code)
result_dir = os.path.join(os.path.dirname(archive_path), disease_code, 'gene_expression')
if not os.path.exists(result_dir):
os.makedirs(result_dir)
# Unpack tar file
archive = tarfile.open(archive_path)
gene_quantification_files = set(el for el in archive if 'genes.normalized_results' in el.name)
if len(gene_quantification_files.difference(os.listdir(result_dir))) > 0:
archive.extractall(members=gene_quantification_files, path=result_dir)
archive.extract('FILE_SAMPLE_MAP.txt', path=result_dir)
# Load map from samples to RNA files
rna_file_sample_map = pd.read_csv(os.path.join(result_dir, 'FILE_SAMPLE_MAP.txt'), sep='\t')
rna_file_sample_map_id_split = rna_file_sample_map['barcode(s)'].str.rsplit('-', n=4, expand=True)
rna_file_sample_map_id_split.columns = ['TCGA_ID', 'SampleID', 'PortionID', 'PlateID', 'CenterID']
rna_file_sample_map = rna_file_sample_map.join(rna_file_sample_map_id_split)
gene_filter = rna_file_sample_map['filename'].str.contains('genes.normalized_results')
gene_rna_file_sample_map = rna_file_sample_map[gene_filter]
rna_dfs = []
for (f, sample) in list(zip(gene_rna_file_sample_map['filename'], gene_rna_file_sample_map['TCGA_ID'])):
sample_rna_df = pd.read_csv(os.path.join(result_dir, f), sep='\t')
sample_rna_df['TCGA_ID'] = sample
sample_rna_df['gene_name'] = sample_rna_df.gene_id.str.split('|').str.get(0)
rna_dfs.append(sample_rna_df)
rna_df = pd.concat(rna_dfs, copy=False).merge(rna_file_sample_map)
if with_clinical:
patient_data_df = load_clinical_data(disease_code)
merged = rna_df.merge(patient_data_df,
how='outer',
left_on='TCGA_ID',
right_on='bcr_patient_barcode')
return merged
else:
return rna_df
def load_mutation_data(disease_code,
with_clinical=False,
variant_type='all',
wait_time=30):
"""Load variants from TCGA
Parameters
----------
disease_code : str
with_clinical : bool, optional
If True, attach the clinical information
variant_type : str, optional
Filter to a specific variant type 'SNP', 'INDEL'
wait_time : int, optional
Time to wait for response from TCGA
Returns
-------
mutations : Pandas dataframe
A dataframe of mutations
"""
archive_path = prefetch_mutation_data(disease_code,
wait_time=wait_time,
cache=True)
# Unpack tar file
archive = tarfile.open(archive_path)
result_dir = os.path.join(os.path.dirname(archive_path), disease_code,
'mutations')
if not os.path.exists(result_dir):
os.makedirs(result_dir)
archive.extractall(path=result_dir)
# Filter to MAF files
maf_files = [f for f in os.listdir(result_dir) if f.endswith('.maf')]
mutation_df = pd.concat([
pd.read_csv(os.path.join(result_dir, maf_file),
sep='\t',
na_values='[Not Available]') for maf_file in maf_files
],
copy=False)
# Expand out the TCGA barcode to retrieve the TCGA ID
tcga_info = mutation_df['Tumor_Sample_Barcode'].str.rsplit('-',
n=4,
expand=True)
tcga_info.columns = [
'TCGA_ID', 'SampleID', 'PortionID', 'PlateID', 'CenterID'
]
mutations = mutation_df.join(tcga_info, how='left')
if variant_type != 'all':
if variant_type == 'indel':
mutations = mutations[(mutations['Variant_Type'] == 'INS') |
(mutations['Variant_Type'] == 'DEL')]
else:
mutations = mutations[mutations['Variant_Type'] == variant_type]
logging.info("Loaded {} mutations for {} tumors from {} patients".format(
len(mutations), mutations['Tumor_Sample_Barcode'].nunique(),
mutations['TCGA_ID'].nunique()))
if with_clinical:
patient_data_df = load_clinical_data(disease_code)
merged = mutations.merge(patient_data_df,
how='outer',
left_on='TCGA_ID',
right_on='bcr_patient_barcode')
logging.info("Patients: {}, Tumor Samples: {}, Mutations {}".format(
merged['bcr_patient_barcode'].nunique(),
merged['Tumor_Sample_Barcode'].nunique(),
len(merged[~merged['TCGA_ID'].isnull()])))
return merged
else:
return mutations
def load_mutation_data(disease_code,
with_clinical=False,
variant_type='all',
wait_time=30):
"""Load variants from TCGA
Parameters
----------
disease_code : str
with_clinical : bool, optional
If True, attach the clinical information
variant_type : str, optional
Filter to a specific variant type 'SNP', 'INDEL'
wait_time : int, optional
Time to wait for response from TCGA
Returns
-------
mutations : Pandas dataframe
A dataframe of mutations
"""
archive_path = prefetch_mutation_data(disease_code,
wait_time=wait_time,
cache=True)
# Unpack tar file
archive = tarfile.open(archive_path)
result_dir = os.path.join(os.path.dirname(archive_path), disease_code, 'mutations')
if not os.path.exists(result_dir):
os.makedirs(result_dir)
archive.extractall(path=result_dir)
# Filter to MAF files
maf_files = [f
for f in os.listdir(result_dir)
if f.endswith('.maf')]
mutation_df = pd.concat([pd.read_csv(os.path.join(result_dir, maf_file),
sep='\t',
na_values='[Not Available]')
for maf_file in maf_files], copy=False)
# Expand out the TCGA barcode to retrieve the TCGA ID
tcga_info = mutation_df['Tumor_Sample_Barcode'].str.rsplit('-', n=4, expand=True)
tcga_info.columns = ['TCGA_ID', 'SampleID', 'PortionID', 'PlateID', 'CenterID']
mutations = mutation_df.join(tcga_info, how='left')
if variant_type != 'all':
if variant_type == 'indel':
mutations = mutations[
(mutations['Variant_Type'] == 'INS') |
(mutations['Variant_Type'] == 'DEL')
]
else:
mutations = mutations[mutations['Variant_Type'] == variant_type]
logging.info("Loaded {} mutations for {} tumors from {} patients".format(
len(mutations),
mutations['Tumor_Sample_Barcode'].nunique(),
mutations['TCGA_ID'].nunique()
)
)
if with_clinical:
patient_data_df = load_clinical_data(disease_code)
merged = mutations.merge(patient_data_df,
how='outer',
left_on='TCGA_ID',
right_on='bcr_patient_barcode')
logging.info("Patients: {}, Tumor Samples: {}, Mutations {}".format(
merged['bcr_patient_barcode'].nunique(),
merged['Tumor_Sample_Barcode'].nunique(),
len(merged[~merged['TCGA_ID'].isnull()])
)
)
return merged
else:
return mutations
