def pull_as_onehot(self, batch_size):
#Calls the pull function to retrieve a list of nucleotide sequences
#Converts the nucleotides to a [batch_size,4,seq_len] onehot numpy array
nuc_list = self.pull(batch_size)
dna_batch = np.zeros((batch_size, 4, self.seq_len))
#labels_batch = np.ones((batch_size))*label
for bi, seq in enumerate(nuc_list):
dna_batch[bi, :, :] = dbt.seq_to_onehot(seq)
return dna_batch
def relevance_from_nucs(self, nucs, label):
"""
Return the relevance of a nucleotide sequence and corresponding label
:nuc_seq: string nucleotide sequence
:label: 1 x num_classes numpy indicator array
:returns: 4xn relevance matrix
:rtype: numpy array
"""
return self.run_relevance(dbt.seq_to_onehot(nuc_seq), label)
def main():
nucs = ['ATACGACGACT','GTACGACGTAC','GTCGATGCATG']
nuc_3d = np.zeros( ( len(nucs),len(nucs[0]),4 ) )
for i,nuc in enumerate(nucs):
nuc_3d[i,:,:] = dbt.seq_to_onehot_topdown(nuc)
nuc_3d_ten = tf.stack(nuc_3d)
#test_reverse_sequence(nuc_3d)
print "Test complement"
#test_com(nuc_3d_ten)
print "Test reverse complement"
#test_revcom(nuc_3d_ten)
#test_nuc_input(nuc_3d_ten)
print "Test seq to onehot"
print dbt.seq_to_onehot('AACGTCG').T
def pull_index_onehot(self, index):
numeric_label = self.label_from_index(index)
iseq = self.seq_parser_list[index].seq
nuc_onehot = dbt.seq_to_onehot(iseq) #(outputs nx4)
return numeric_label, nuc_onehot
def pull_index_onehot(self, index):
return self.labels[index], dbt.seq_to_onehot(self.nuc_seqs[index])