def process_log(muted, sims_dir= ''):
'''
verify directories indicated in log exist in given directory.
'''
available= get_available_muts(muted)
### cleaning data set
### i.e. accounting for aborted runs.
available, miss_data= check_availability(available, dir_check=sims_dir)
available, empty= clean_empty(available,str_format= '',dir_check= sims_dir,requested= ['.vcf.gz'])
return available
def process_dir(sims_dir= ''):
'''
verify directories indicated in log exist in given directory.
'''
available= [name for name in os.listdir(sims_dir)]
### cleaning data set
### i.e. accounting for aborted runs.
available, miss_data= check_availability(available, dir_check=sims_dir)
available, empty= clean_empty(available,str_format= '',dir_check= sims_dir,requested= ['.vcf.gz','fa.gz'])
print('missing: {}, no vcf: {}'.format(len(miss_data),len(empty)))
return available