def __plink_to_npz(self, p, file_name):
'''Convert p from plink to npz format using the file set specified by file_name.'''
npz = file_name+'.npz'
# Save test problem in plink format
io.write_plink(p, prefix=file_name)
# Convert plink -> npz
io.plink_to_npz(file_name, npz)
# Load npz and check that the problem object didn't change
p2 = io.read_npz(npz)
return p2
def __save_and_load_problem_plink(self, problem):
'''Save and load a problem from PLINK file set.'''
try:
# Get a temporary file name
f = tempfile.NamedTemporaryFile(delete=False)
file_name = f.name
f.close()
io.write_plink(problem, file_name)
return io.read_plink(prefix=file_name)
finally:
# Delete test files
for ext in ['', '.pdg.tfam', '.tfam', '.tped', '.hap.tped', '.info']:
os.remove(file_name + ext)
def __main(options):
'''
--------------------------------------------------
Main program - accepts an options struct.
--------------------------------------------------
'''
if options.debug: print 'Input options', options
print 'Building phaser (stage = %d) ...' % (options.stage, )
phaser = build_phasing_pipeline(options)
if options.debug: print 'Reading data ...'
problem = __load_problem(options)
if options.debug: print 'Phasing ...'
params = PhaseParam()
params.update_from_struct(options)
request = run_phasing_chain(phaser, problem, params)
print ''
request.stats.pprint()
print ''
if options.output is not None:
if options.min_output:
print 'Minimizing output size...'
io.slim(problem)
out_prefix, ext = os.path.splitext(options.output)
if ext == '.npz':
print 'Writing haplotype result to %s in NPZ format ...' % (
options.output, )
io.write_npz(problem, options.output)
output_info = out_prefix + '.info.npz'
print 'Writing problem info result to %s in NPZ format ...' % (
output_info, )
io.write_info_npz(problem.info, output_info)
else:
print 'Writing haplotype result to %s in PLINK format ...' % (
options.output, )
io.write_plink(problem, options.output, verbose=options.debug)
return problem
def __main(options):
'''
--------------------------------------------------
Main program - accepts an options struct.
--------------------------------------------------
'''
if options.debug: print 'Input options', options
print 'Building phaser (stage = %d) ...' % (options.stage,)
phaser = build_phasing_pipeline(options)
if options.debug: print 'Reading data ...'
problem = __load_problem(options)
if options.debug: print 'Phasing ...'
params = PhaseParam()
params.update_from_struct(options)
request = run_phasing_chain(phaser, problem, params)
print ''
request.stats.pprint()
print ''
if options.output is not None:
if options.min_output:
print 'Minimizing output size...'
io.slim(problem)
out_prefix, ext = os.path.splitext(options.output)
if ext == '.npz':
print 'Writing haplotype result to %s in NPZ format ...' % (options.output,)
io.write_npz(problem, options.output)
output_info = out_prefix + '.info.npz'
print 'Writing problem info result to %s in NPZ format ...' % (output_info,)
io.write_info_npz(problem.info, output_info)
else:
print 'Writing haplotype result to %s in PLINK format ...' % (options.output,)
io.write_plink(problem, options.output, verbose=options.debug)
return problem
# Convert plink tped -> npz
problem = io.read_plink(prefix=base_name, pedigree=pedigree_file, haplotype=None,
verbose=options.debug)
# Phase, impute, fill missing
phaser = phase.build_phasing_pipeline(options)
request = phase.run_phasing_chain(phaser, problem)
stats = request.stats
print ''
stats.pprint()
print ''
# Convert phased npz -> plink tped. Save only genotypes (haplotypes may need to be saved in the stats
# object as a hash table for 'coloring the pedigree' at a later stage.
genotype_file = out_base_name + '.tped'
io.write_plink(problem, out_base_name, verbose=True,
save_node_type=False, save_genotype=True, save_haplotype=False, save_error=False)
# Save statistics and phasing metadata in a separate npz
np.savez(out_base_name + '.stats',
stats=np.array([stats]),
info=np.array([problem.info]),
pedigree=np.array([problem.pedigree]))
plink_cmd_base = '%s --tfile %s' % (bu.PLINK, out_base_name,)
if options.recode:
# First, compute allele frequencies with PLINK
util.run_command('%s --nonfounders --freq --out %s' % (plink_cmd_base, out_base_name))
# Convert frequencies file that to a reference allele recoding
# file (a file containing the list of SNPs and their minor allele letter)
bu.frq_to_minor_file(out_base_name + '.frq', out_base_name + '.mnr')
