def test_no_context_constructor():
# no context
gs = GeneSequence(gene_fa, nuc_context=0)
gs.set_gene(bed)
sc = SequenceContext(gs)
true_counts = {'None': 21}
true_ctxt2pos = {'None': range(21)}
_check_true_counts(sc, true_counts)
_check_true_context_pos(sc, true_ctxt2pos)
def test_single_context_constructor():
# single nuc context
gs = GeneSequence(gene_fa, nuc_context=1)
gs.set_gene(bed)
sc = SequenceContext(gs)
true_counts = {'A': 10,
'T': 4,
'G': 4,
'C': 3}
true_ctxt2pos = {'A': [0, 2, 5, 6, 9, 11, 13, 18, 19, 20],
'C': [1, 15, 16],
'G': [4, 8, 12, 17],
'T': [3, 7, 10, 14]}
_check_true_counts(sc, true_counts)
_check_true_context_pos(sc, true_ctxt2pos)
def test_chasm_context_constructor():
# chasm context, mixture between single and di context
gs = GeneSequence(gene_fa, nuc_context=1.5)
gs.set_gene(bed)
sc = SequenceContext(gs)
true_counts = {'A': 10,
'C': 1,
'C*pG': 1,
'CpG*': 1,
'G*pA': 3,
'T': 4,
'TpC*': 1}
true_ctxt2pos = {'A': [2, 5, 6, 9, 11, 13, 18, 19, 0, 20],
'C': [1],
'C*pG': [16],
'CpG*': [17],
'G*pA': [4, 8, 12],
'T': [3, 7, 10, 14],
'TpC*': [15]}
_check_true_counts(sc, true_counts)
_check_true_context_pos(sc, true_ctxt2pos)
def singleprocess_permutation(info):
# initialize input
bed_list, mut_df, opts, fs_cts_df, p_inactivating = info
current_chrom = bed_list[0].chrom
logger.info('Working on chromosome: {0} . . .'.format(current_chrom))
gene_fa = pysam.Fastafile(opts['input'])
gs = GeneSequence(gene_fa, nuc_context=opts['context'])
# list of columns that are needed
cols = [
'Chromosome',
'Start_Position',
'Reference_Allele',
'Tumor_Allele',
'Variant_Classification',
]
# conditionally add protein_change column if exists
if 'Protein_Change' in mut_df.columns:
cols += ['Protein_Change']
# figure out which genes actually have a mutation
genes_with_mut = set(mut_df['Gene'].unique())
# iterate through each gene
result = []
for bed in bed_list:
if bed.gene_name not in genes_with_mut:
# skip genes with no mutations
continue
# prepare info for running permutation test
mut_info = mut_df.loc[mut_df['Gene'] == bed.gene_name, cols]
gs.set_gene(bed)
sc = SequenceContext(gs, seed=opts['seed'])
# count total mutations in gene
total_mut = len(mut_info)
# fix nucleotide letter if gene is on - strand
if bed.strand == '-':
rc = mut_info['Tumor_Allele'].map(lambda x: utils.rev_comp(x))
mut_info.loc[:, 'Tumor_Allele'] = rc
# get coding positions, mutations unmapped to the reference tx will have
# NA for a coding position
pos_list = []
for ix, row in mut_info.iterrows():
coding_pos = bed.query_position(bed.strand, row['Chromosome'],
row['Start_Position'])
pos_list.append(coding_pos)
mut_info.loc[:, 'Coding Position'] = pos_list
# recover mutations that could not be mapped to the reference transcript
# for a gene before being dropped (next step)
unmapped_mut_info = mc.recover_unmapped_mut_info(
mut_info, bed, sc, opts)
# drop mutations wich do not map to reference tx
mut_info = mut_info.dropna(subset=['Coding Position'
]) # mutations need to map to tx
mut_info['Coding Position'] = mut_info['Coding Position'].astype(int)
num_mapped_muts = len(mut_info)
unmapped_muts = total_mut - num_mapped_muts
# construct sequence context
#gs.add_germline_variants(mut_info['Reference_Allele'].tolist(),
# mut_info['Coding Position'].tolist())
# calculate results of permutation test
if opts['kind'] == 'oncogene':
# calculate position based permutation results
tmp_result = mypval.calc_position_p_value(
mut_info,
unmapped_mut_info,
sc,
gs,
bed,
opts['score_dir'],
opts['num_iterations'],
opts['stop_criteria'],
0, # no recurrent mutation pseudo count
opts['recurrent'],
opts['fraction'])
result.append(tmp_result + [total_mut, unmapped_muts])
elif opts['kind'] == 'tsg':
# calculate results for deleterious mutation permutation test
#fs_ct = fs_cts_df['total'][bed.gene_name]
#fs_unmapped = fs_cts_df['unmapped'][bed.gene_name]
# replaced fs_ct with zero to stop using the frameshifts in
# simulation
tmp_result = mypval.calc_deleterious_p_value(
mut_info,
unmapped_mut_info,
sc,
gs,
bed,
opts['num_iterations'],
opts['stop_criteria'],
opts['deleterious'],
0, # no deleterious mutation pseudo count
opts['seed'])
result.append(tmp_result + [num_mapped_muts, unmapped_muts])
#fs_ct, fs_unmapped])
elif opts['kind'] == 'hotmaps1d':
# save null distribution if user option specified
if opts['null_distr_dir']:
if not os.path.exists(opts['null_distr_dir']):
os.mkdir(opts['null_distr_dir'])
save_path = os.path.join(opts['null_distr_dir'],
bed.gene_name + '.{0}.txt')
else:
save_path = None
# calculate position based permutation results
mywindow = list(map(int, opts['window'].split(',')))
tmp_result = mypval.calc_hotmaps_p_value(mut_info,
unmapped_mut_info,
sc,
gs,
bed,
mywindow,
opts['num_iterations'],
opts['stop_criteria'],
opts['report_index'],
null_save_path=save_path)
result.extend(tmp_result)
elif opts['kind'] == 'protein':
tmp_result = mypval.calc_protein_p_value(
mut_info, unmapped_mut_info, sc, gs, bed,
opts['neighbor_graph_dir'], opts['num_iterations'],
opts['stop_criteria'], opts['recurrent'], opts['fraction'])
result.append(tmp_result + [total_mut, unmapped_muts])
else:
# calc results for entropy-on-effect permutation test
tmp_result = mypval.calc_effect_p_value(
mut_info,
unmapped_mut_info,
sc,
gs,
bed,
opts['num_iterations'],
0, # no recurrent mutation pseudo count
opts['recurrent'],
opts['fraction'])
result.append(tmp_result + [total_mut, unmapped_muts])
gene_fa.close()
logger.info('Finished working on chromosome: {0}.'.format(current_chrom))
return result
def test_dinuc_context_constructor():
# dinucleotide context
gs = GeneSequence(gene_fa, nuc_context=2)
gs.set_gene(bed)
sc = SequenceContext(gs)