def insert_sequence_pos(pan_sequence, second_sequence):
sequence_id = ''
#check if blocks exist with the given position in pan genome, strain
#get block from strain position
block_strain = build_block(None, second_sequence.start, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
#get block from pan position
block_pan = build_block(pan_sequence.start, None, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
block_encase_pan = encase_block(pan_sequence.start, None, pan_sequence.end, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
block_encase_strain = encase_block(None, second_sequence.start, second_sequence.end, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
if (block_encase_pan != None) or (block_encase_strain != None):
print "encase %s, %s, %s, %s, %s" % (pan_sequence.start, second_sequence.start, second_sequence.length, pan_sequence.length, pan_sequence.sequence_string)
elif (block_strain == None) and (block_pan == None):
sequence_id = insert_sequence(pan_sequence, second_sequence)
insert_position(pan_sequence.start, second_sequence.start, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name, sequence_id)
#map end from pan -> strain
check_end_block_surrounding(pan_sequence, second_sequence)
#map end from strain -> pan
check_end_block_surrounding_strain(pan_sequence, second_sequence)
elif (block_strain.length == 0) and (block_pan.length == 0):
sequence_id = insert_sequence(pan_sequence, second_sequence)
insert_position(pan_sequence.start, second_sequence.start, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name, sequence_id)
#map end from pan -> strain
check_end_block_surrounding(pan_sequence, second_sequence)
#map end from strain -> pan
check_end_block_surrounding_strain(pan_sequence, second_sequence)
def check_end_block_surrounding_strain(pan_sequence, second_sequence):
start = second_sequence.end + 1
total_length = int(second_sequence.total_length)
if start < total_length:
start_block = position_end_block(None, start, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
if start_block == None:
insert_position('-1', start, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
#the end of the sequence is reached
elif start >= total_length:
start_block = position_end_block(None, start, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
if start_block == None:
#'-1' should we mark the end of a sequence differently?
insert_position('-1', start, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
def check_mapping_sequence_start(pan_sequence, second_sequence):
block_position_start = position_exist(0, None, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
if block_position_start == None:
insert_position('0', '-1', second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
def check_mapping_sequence_end(pan_sequence, second_sequence):
last_position = int(pan_sequence.total_length) + 1
block_position_end = position_end_block(last_position, None, second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
if block_position_end == None:
insert_position(last_position, '-1', second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)
def check_start_block_surrounding(pan_sequence, second_sequence):
start = int(pan_sequence.start) - 1
if start > 0:
start_block = position_start_block(start, None, second_sequence.strain_name, pan_sequence.sequence_name, second_sequence.sequence_name)
if start_block == None:
insert_position(start, '-2', second_sequence.strain_name, second_sequence.sequence_name, pan_sequence.sequence_name)