def addTestFile(self, field='ind_id', file_pat='/mnt/scratch/asalomatov/data/SSC/wes/feature_sets/fb/all_SNP/%s'): num_subst = len(re.findall('\%s', file_pat)) print('%s substitutions found' % num_subst) if num_subst > 0: x = self.ped[field].apply(lambda f: func.listFiles(file_pat % ((f,) * num_subst))) self.ped['test'] = pd.Series(x, index=self.ped.index) else: self.ped['test'] = file_pat
def addBai(self, field='ind_id', file_pat='/mnt/ceph/asalomatov/SSC_Eichler/data_S3/%s*bam.bai'): num_subst = len(re.findall('\%s', file_pat)) print('%s substitutions found' % num_subst) if num_subst > 0: x = self.ped[field].apply(lambda f: func.listFiles(file_pat % ((f,) * num_subst))) self.ped['bai'] = pd.Series(x, index=self.ped.index) else: self.ped['bai'] = file_pat
def addBai(self, field='ind_id', file_pat='/mnt/ceph/asalomatov/SSC_Eichler/data_S3/%s*bam.bai'): num_subst = len(re.findall('\%s', file_pat)) print('%s substitutions found' % num_subst) if num_subst > 0: x = self.ped[field].apply( lambda f: func.listFiles(file_pat % ((f, ) * num_subst))) self.ped['bai'] = pd.Series(x, index=self.ped.index) else: self.ped['bai'] = file_pat
def addTestFile( self, field='ind_id', file_pat='/mnt/scratch/asalomatov/data/SSC/wes/feature_sets/fb/all_SNP/%s' ): num_subst = len(re.findall('\%s', file_pat)) print('%s substitutions found' % num_subst) if num_subst > 0: x = self.ped[field].apply( lambda f: func.listFiles(file_pat % ((f, ) * num_subst))) self.ped['test'] = pd.Series(x, index=self.ped.index) else: self.ped['test'] = file_pat
def addBam(self, field='ind_id', file_pat='/mnt/ceph/asalomatov/SSC_Eichler/data_S3/%s*.bam', num_subst=2): """Sometimes it is easier to add a bam column using a sting substitution pattern. Usually it is done manually""" num_subst = len(findall('\%s', file_pat)) print num_subst, ' substitutions found' if num_subst > 0: x = self.ped[field].apply( lambda f: func.listFiles(file_pat % ((f, ) * num_subst))) self.ped['bam'] = pd.Series(x, index=self.ped.index) else: self.ped['bam'] = file_pat