def main(opts):
# read in mutations
mut_df = pd.read_csv(opts['mutations'], sep='\t')
orig_num_mut = len(mut_df)
# correct chromosome names
mut_df['Chromosome'] = correct_chrom_names(mut_df['Chromosome'])
# fix additional issues
mut_df = mut_df.dropna(subset=['Tumor_Allele', 'Start_Position', 'Chromosome'])
logger.info('Kept {0} mutations after droping mutations with missing '
'information (Droped: {1})'.format(len(mut_df), orig_num_mut - len(mut_df)))
mut_df = utils._fix_mutation_df(mut_df)
# read genome fasta file
genome_fa = pysam.Fastafile(opts['fasta'])
# read BED file for transcripts
bed_dict = utils.read_bed(opts['bed'], [])
gene2bed = {item.gene_name: item
for bed_list in bed_dict.values()
for item in bed_list}
# group mutations by gene
mut_grpby = mut_df.groupby('Gene')
unmapped_mut_list = []
for i, mut_info in mut_grpby:
gene_name = mut_info['Gene'].iloc[0]
# try to find tx annotation for gene
bed = None
try:
bed = gene2bed[gene_name]
except KeyError:
pass
if bed:
# get coding positions, mutations unmapped to the reference tx will have
# NA for a coding position
for ix, row in mut_info.iterrows():
coding_pos = bed.query_position(bed.strand,
row['Chromosome'],
row['Start_Position'])
if not coding_pos:
unmapped_mut_list.append(row.tolist())
else:
#unmapped_mut_df = pd.concat([unmapped_mut_df, mut_info])
unmapped_mut_list += mut_info.values.tolist()
# save the unmapped mutations to a file
unmapped_mut_df = pd.DataFrame(unmapped_mut_list, columns=mut_df.columns)
logger.info('{0} mutations were unmappable to a '
'reference transcript'.format(len(unmapped_mut_df)))
unmapped_mut_df.to_csv(opts['unmapped'], sep='\t', index=False)
coord_base, coord_strand = detect_coordinates(mut_df, genome_fa)
logger.info('RESULT: {0}-based coordinates, positions reported on {1} strand'.format(coord_base, coord_strand))
genome_fa.close()
def main(opts):
# hack to index the FASTA file
gene_fa = pysam.Fastafile(opts['input'])
gene_fa.close()
# Get Mutations
mut_df = pd.read_csv(opts['mutations'], sep='\t')
orig_num_mut = len(mut_df)
# rename columns to fit my internal column names
rename_dict = {
'Hugo_Symbol': 'Gene',
'Tumor_Sample_Barcode': 'Tumor_Sample',
'Tumor_Seq_Allele2' : 'Tumor_Allele'
}
mut_df.rename(columns=rename_dict, inplace=True)
# restrict to only observed genes if flag present
restricted_genes = None
if opts['restrict_genes']:
restricted_genes = set(mut_df['Gene'].unique())
# process indels
indel_df = indel.keep_indels(mut_df) # return indels only
indel_df.loc[:, 'Start_Position'] = indel_df['Start_Position'] - 1 # convert to 0-based
indel_df.loc[:, 'indel len'] = indel_df['indel len'] + 1
logger.info('There were {0} indels identified.'.format(len(indel_df)))
mut_df = mut_df.dropna(subset=['Tumor_Allele', 'Start_Position', 'Chromosome'])
logger.info('Kept {0} mutations after droping mutations with missing '
'information (Droped: {1})'.format(len(mut_df), orig_num_mut - len(mut_df)))
# select valid single nucleotide variants only
mut_df = utils._fix_mutation_df(mut_df, opts['unique'])
# read in bed info
bed_dict = utils.read_bed(opts['bed'], restricted_genes)
# perform permutation
opts['handle'] = open(opts['output'], 'w')
multiprocess_permutation(bed_dict, mut_df, opts, indel_df)
# save indels
if opts['maf']:
#with open(opts['output'], 'a') as handle:
mywriter = csv.writer(opts['handle'], delimiter='\t', lineterminator='\n')
for maf_lines in indel.simulate_indel_maf(indel_df, bed_dict,
opts['num_iterations'],
opts['seed']):
mywriter.writerows(maf_lines)
opts['handle'].close()
def main(opts, mut_df=None, frameshift_df=None):
# hack to index the FASTA file
gene_fa = pysam.Fastafile(opts['input'])
gene_fa.close()
# Get Mutations
if mut_df is None:
mut_df = pd.read_csv(opts['mutations'], sep='\t')
orig_num_mut = len(mut_df)
# rename columns to fit my internal column names
rename_dict = {
'Hugo_Symbol': 'Gene',
'Tumor_Sample_Barcode': 'Tumor_Sample',
'Tumor_Seq_Allele2': 'Tumor_Allele'
}
mut_df.rename(columns=rename_dict, inplace=True)
# drop rows with missing info
na_cols = ['Gene', 'Tumor_Allele', 'Start_Position', 'Chromosome']
mut_df = mut_df.dropna(subset=na_cols)
logger.info('Kept {0} mutations after droping mutations with missing '
'information (Droped: {1})'.format(len(mut_df),
orig_num_mut - len(mut_df)))
# count frameshifts
if opts['kind'] == 'tsg':
if frameshift_df is None:
# read in mutations
if mut_df is None:
mut_df = pd.read_csv(opts['mutations'], sep='\t')
# count number of frameshifts
frameshift_df = cf.count_frameshift_total(mut_df, opts['bed'],
opts['use_unmapped'])
# calculate the proportion of inactivating
#num_inact = len(mut_df[mut_df['Variant_Classification'].isin(utils.variant_inactivating)])
#num_non_inact = len(mut_df[mut_df['Variant_Classification'].isin(utils.variant_non_inactivating)])
num_fs = len(mut_df[mut_df['Variant_Classification'].isin(
utils.variant_frameshift)])
num_all = len(mut_df[mut_df['Variant_Classification'].isin(
utils.all_variants)])
#p_inactivating = float(num_inact) / (num_inact + num_non_inact)
p_inactivating = float(num_fs) / num_all
# select valid single nucleotide variants only
mut_df = utils._fix_mutation_df(mut_df, opts['unique'])
# log random number seed choice if provided
if opts['seed'] is not None:
logger.info('Pseudo Random Number Generator Seed: {0}'.format(
opts['seed']))
# read BED file
bed_dict = utils.read_bed(opts['bed'])
# Perform BH p-value adjustment and tidy up data for output
if opts['kind'] == 'oncogene':
permutation_result = multiprocess_permutation(bed_dict, mut_df, opts)
permutation_df = pr.handle_oncogene_results(permutation_result,
opts['num_iterations'])
elif opts['kind'] == 'tsg':
permutation_result = multiprocess_permutation(bed_dict, mut_df, opts,
frameshift_df,
p_inactivating)
permutation_df = pr.handle_tsg_results(permutation_result)
elif opts['kind'] == 'hotmaps1d':
permutation_result = multiprocess_permutation(bed_dict, mut_df, opts)
#frameshift_df, p_inactivating)
permutation_df = pr.handle_hotmaps_results(permutation_result)
elif opts['kind'] == 'protein':
permutation_result = multiprocess_permutation(bed_dict, mut_df, opts)
permutation_df = pr.handle_protein_results(permutation_result)
elif opts['kind'] == 'effect':
permutation_result = multiprocess_permutation(bed_dict, mut_df, opts)
permutation_df = pr.handle_effect_results(permutation_result)
# save output
if opts['output']:
permutation_df.to_csv(opts['output'], sep='\t', index=False)
return permutation_df
def main(opts):
global cols
if opts['score_dir']:
cols.extend(['Total MGAEntropy', 'Total Missense VEST'])
# hack to index the FASTA file
gene_fa = pysam.Fastafile(opts['input'])
gene_fa.close()
# Get Mutations
mut_df = pd.read_csv(opts['mutations'], sep='\t')
orig_num_mut = len(mut_df)
mut_df = mut_df.dropna(
subset=['Tumor_Allele', 'Start_Position', 'Chromosome'])
logger.info('Kept {0} mutations after droping mutations with missing '
'information (Droped: {1})'.format(len(mut_df),
orig_num_mut - len(mut_df)))
# select valid single nucleotide variants only
mut_df = utils._fix_mutation_df(mut_df)
# read in bed info
bed_dict = utils.read_bed(opts['bed'], [])
# perform permutation test
#permutation_result = multiprocess_permutation(bed_dict, mut_df, opts)
sim_result, obs_result = multiprocess_permutation(bed_dict, mut_df, opts)
# report number of observed non-silent and silent mutations
#obs_result = [x[1] for x in permutation_result] # actually observed num mutations
#obs_result = permutation_result[1] # actually observed num mutations
if not opts['by_sample']:
total_non_silent = sum(o[0] for o in obs_result)
total_silent = sum(o[1] for o in obs_result)
total_nonsense = sum(o[2] for o in obs_result)
total_loststop = sum(o[3] for o in obs_result)
total_splice_site = sum(o[4] for o in obs_result)
total_loststart = sum(o[5] for o in obs_result)
total_missense = sum(o[6] for o in obs_result)
if opts['score_dir']:
total_mgaentropy = sum(o[7] for o in obs_result)
total_vest = sum(o[8] for o in obs_result)
logger.info(
'There were {0} non-silent SNVs and {1} silent SNVs actually '
'observed from the provided mutations.'.format(
total_non_silent, total_silent))
logger.info(
'There were {0} missense SNVs, {1} nonsense SNVs, {2} lost stop SNVs, '
', {3} lost start, and {4} splice site SNVs'.format(
total_missense, total_nonsense, total_loststop,
total_loststart, total_splice_site))
else:
obs_non_silent_df = obs_result
#sim_result = [s[0] for s in permutation_result] # results with permutation
#sim_result = permutation_result[0]
# convert to dataframe to save to file
non_silent_ratio_df = pd.DataFrame(sim_result, columns=cols)
# save simulation output
non_silent_ratio_df.to_csv(opts['output'], sep='\t', index=False)
# save observed values if file provided
if opts['observed_output']:
if not opts['by_sample']:
obs_result = [
total_non_silent, total_silent, total_nonsense, total_loststop,
total_splice_site, total_loststart, total_missense
]
if opts['score_dir']:
obs_result.extend([total_mgaentropy, total_vest])
obs_non_silent_df = pd.DataFrame([obs_result], columns=cols)
obs_non_silent_df.to_csv(opts['observed_output'],
sep='\t',
index=False)
else:
obs_non_silent_df.to_csv(opts['observed_output'], sep='\t')
return non_silent_ratio_df
def main(opts):
# read in mutations
mut_df = pd.read_csv(opts['mutations'], sep='\t')
orig_num_mut = len(mut_df)
# correct chromosome names
mut_df['Chromosome'] = correct_chrom_names(mut_df['Chromosome'])
# fix additional issues
mut_df = mut_df.dropna(
subset=['Tumor_Allele', 'Start_Position', 'Chromosome'])
logger.info('Kept {0} mutations after droping mutations with missing '
'information (Droped: {1})'.format(len(mut_df),
orig_num_mut - len(mut_df)))
mut_df = utils._fix_mutation_df(mut_df)
# read genome fasta file
genome_fa = pysam.Fastafile(opts['fasta'])
# read BED file for transcripts
bed_dict = utils.read_bed(opts['bed'], [])
gene2bed = {
item.gene_name: item
for bed_list in bed_dict.values() for item in bed_list
}
# group mutations by gene
mut_grpby = mut_df.groupby('Gene')
unmapped_mut_list = []
for i, mut_info in mut_grpby:
gene_name = mut_info['Gene'].iloc[0]
# try to find tx annotation for gene
bed = None
try:
bed = gene2bed[gene_name]
except KeyError:
pass
if bed:
# get coding positions, mutations unmapped to the reference tx will have
# NA for a coding position
for ix, row in mut_info.iterrows():
coding_pos = bed.query_position(bed.strand, row['Chromosome'],
row['Start_Position'])
if not coding_pos:
unmapped_mut_list.append(row.tolist())
else:
#unmapped_mut_df = pd.concat([unmapped_mut_df, mut_info])
unmapped_mut_list += mut_info.values.tolist()
# save the unmapped mutations to a file
unmapped_mut_df = pd.DataFrame(unmapped_mut_list, columns=mut_df.columns)
logger.info('{0} mutations were unmappable to a '
'reference transcript'.format(len(unmapped_mut_df)))
unmapped_mut_df.to_csv(opts['unmapped'], sep='\t', index=False)
coord_base, coord_strand = detect_coordinates(mut_df, genome_fa)
logger.info(
'RESULT: {0}-based coordinates, positions reported on {1} strand'.
format(coord_base, coord_strand))
genome_fa.close()
def main(opts):
global cols
if opts['score_dir']:
cols.extend(['Total MGAEntropy', 'Total Missense VEST'])
# hack to index the FASTA file
gene_fa = pysam.Fastafile(opts['input'])
gene_fa.close()
# Get Mutations
mut_df = pd.read_csv(opts['mutations'], sep='\t')
orig_num_mut = len(mut_df)
mut_df = mut_df.dropna(subset=['Tumor_Allele', 'Start_Position', 'Chromosome'])
logger.info('Kept {0} mutations after droping mutations with missing '
'information (Droped: {1})'.format(len(mut_df), orig_num_mut - len(mut_df)))
# select valid single nucleotide variants only
mut_df = utils._fix_mutation_df(mut_df)
# read in bed info
bed_dict = utils.read_bed(opts['bed'])
# perform permutation test
#permutation_result = multiprocess_permutation(bed_dict, mut_df, opts)
sim_result, obs_result = multiprocess_permutation(bed_dict, mut_df, opts)
# report number of observed non-silent and silent mutations
#obs_result = [x[1] for x in permutation_result] # actually observed num mutations
#obs_result = permutation_result[1] # actually observed num mutations
if not opts['by_sample']:
total_non_silent = sum(o[0] for o in obs_result)
total_silent = sum(o[1] for o in obs_result)
total_nonsense = sum(o[2] for o in obs_result)
total_loststop = sum(o[3] for o in obs_result)
total_splice_site = sum(o[4] for o in obs_result)
total_loststart = sum(o[5] for o in obs_result)
total_missense = sum(o[6] for o in obs_result)
if opts['score_dir']:
total_mgaentropy = sum(o[7] for o in obs_result)
total_vest = sum(o[8] for o in obs_result)
logger.info('There were {0} non-silent SNVs and {1} silent SNVs actually '
'observed from the provided mutations.'.format(total_non_silent,
total_silent))
logger.info('There were {0} missense SNVs, {1} nonsense SNVs, {2} lost stop SNVs, '
', {3} lost start, and {4} splice site SNVs'.format(total_missense,
total_nonsense,
total_loststop,
total_loststart,
total_splice_site))
else:
obs_non_silent_df = obs_result
#sim_result = [s[0] for s in permutation_result] # results with permutation
#sim_result = permutation_result[0]
# convert to dataframe to save to file
non_silent_ratio_df = pd.DataFrame(sim_result,
columns=cols)
# save simulation output
non_silent_ratio_df.to_csv(opts['output'], sep='\t', index=False)
# save observed values if file provided
if opts['observed_output']:
if not opts['by_sample']:
obs_result = [total_non_silent, total_silent, total_nonsense,
total_loststop, total_splice_site, total_loststart,
total_missense]
if opts['score_dir']:
obs_result.extend([total_mgaentropy, total_vest])
obs_non_silent_df = pd.DataFrame([obs_result], columns=cols)
obs_non_silent_df.to_csv(opts['observed_output'], sep='\t', index=False)
else:
obs_non_silent_df.to_csv(opts['observed_output'], sep='\t')
return non_silent_ratio_df
def main(opts, mut_df=None, frameshift_df=None):
# hack to index the FASTA file
gene_fa = pysam.Fastafile(opts['input'])
gene_fa.close()
# Get Mutations
if mut_df is None:
mut_df = pd.read_csv(opts['mutations'], sep='\t')
orig_num_mut = len(mut_df)
# rename columns to fit my internal column names
rename_dict = {
'Hugo_Symbol': 'Gene',
'Tumor_Sample_Barcode': 'Tumor_Sample',
'Tumor_Seq_Allele2' : 'Tumor_Allele'
}
mut_df.rename(columns=rename_dict, inplace=True)
# drop rows with missing info
na_cols = ['Gene', 'Tumor_Allele', 'Start_Position', 'Chromosome']
mut_df = mut_df.dropna(subset=na_cols)
logger.info('Kept {0} mutations after droping mutations with missing '
'information (Droped: {1})'.format(len(mut_df), orig_num_mut - len(mut_df)))
# count frameshifts
if opts['kind'] == 'tsg':
if frameshift_df is None:
# read in mutations
if mut_df is None:
mut_df = pd.read_csv(opts['mutations'], sep='\t')
# count number of frameshifts
frameshift_df = cf.count_frameshift_total(mut_df, opts['bed'],
opts['use_unmapped'])
# calculate the proportion of inactivating
#num_inact = len(mut_df[mut_df['Variant_Classification'].isin(utils.variant_inactivating)])
#num_non_inact = len(mut_df[mut_df['Variant_Classification'].isin(utils.variant_non_inactivating)])
num_fs = len(mut_df[mut_df['Variant_Classification'].isin(utils.variant_frameshift)])
num_all = len(mut_df[mut_df['Variant_Classification'].isin(utils.all_variants)])
#p_inactivating = float(num_inact) / (num_inact + num_non_inact)
p_inactivating = float(num_fs) / num_all
# select valid single nucleotide variants only
mut_df = utils._fix_mutation_df(mut_df, opts['unique'])
# log random number seed choice if provided
if opts['seed'] is not None:
logger.info('Pseudo Random Number Generator Seed: {0}'.format(opts['seed']))
# read BED file
bed_dict = utils.read_bed(opts['bed'])
# Perform BH p-value adjustment and tidy up data for output
if opts['kind'] == 'oncogene':
permutation_result = multiprocess_permutation(bed_dict, mut_df, opts)
permutation_df = pr.handle_oncogene_results(permutation_result,
opts['num_iterations'])
elif opts['kind'] == 'tsg':
permutation_result = multiprocess_permutation(bed_dict, mut_df, opts,
frameshift_df, p_inactivating)
permutation_df = pr.handle_tsg_results(permutation_result)
elif opts['kind'] == 'hotmaps1d':
permutation_result = multiprocess_permutation(bed_dict, mut_df, opts)
#frameshift_df, p_inactivating)
permutation_df = pr.handle_hotmaps_results(permutation_result)
elif opts['kind'] == 'protein':
permutation_result = multiprocess_permutation(bed_dict, mut_df, opts)
permutation_df = pr.handle_protein_results(permutation_result)
elif opts['kind'] == 'effect':
permutation_result = multiprocess_permutation(bed_dict, mut_df, opts)
permutation_df = pr.handle_effect_results(permutation_result)
# save output
if opts['output']:
permutation_df.to_csv(opts['output'], sep='\t', index=False)
return permutation_df
