def get_align_matrix_from_fname(self, fname):
"""
2007-03-29
2007-11-05
add comment for this function:
first_block(the reference genome) is used to make sure positions with '-' in the reference would be skipped.
these '-' don't belong to the reference and they have no corresponding position. if they are not taken out, they
would cause positions to be offseted.
"""
inf = open(fname)
iter = fasta_block_iterator(inf)
first_block = iter.next()
positions_to_be_checked_ls = self.get_positions_to_be_checked_ls(
first_block)
align_matrix = []
abbr_name_ls = []
for block in iter:
block = cStringIO.StringIO(block)
header_line = block.readline()
abbr_name_ls.append(header_line[1:-1])
sequence = ''
for line in block:
sequence += line[:-1]
align_one_seq = []
for i in positions_to_be_checked_ls:
align_one_seq.append(nt2number[sequence[i]])
align_matrix.append(align_one_seq)
del iter
del inf
return abbr_name_ls, align_matrix
def get_align_matrix_from_fname(self, fname): """ 2007-03-29 2007-11-05 add comment for this function: first_block(the reference genome) is used to make sure positions with '-' in the reference would be skipped. these '-' don't belong to the reference and they have no corresponding position. if they are not taken out, they would cause positions to be offseted. """ inf = open(fname) iter = fasta_block_iterator(inf) first_block = iter.next() positions_to_be_checked_ls = self.get_positions_to_be_checked_ls(first_block) align_matrix = [] abbr_name_ls = [] for block in iter: block = cStringIO.StringIO(block) header_line = block.readline() abbr_name_ls.append(header_line[1:-1]) sequence = '' for line in block: sequence += line[:-1] align_one_seq = [] for i in positions_to_be_checked_ls: align_one_seq.append(nt2number[sequence[i]]) align_matrix.append(align_one_seq) del iter del inf return abbr_name_ls, align_matrix
def get_align_length_from_fname(self, fname): """ 2007-03-29 """ inf = open(fname) iter = fasta_block_iterator(inf) first_block = iter.next() positions_to_be_checked_ls = self.get_positions_to_be_checked_ls(first_block) del iter del inf return len(positions_to_be_checked_ls)
def get_align_length_from_fname(self, fname):
"""
2007-03-29
"""
inf = open(fname)
iter = fasta_block_iterator(inf)
first_block = iter.next()
positions_to_be_checked_ls = self.get_positions_to_be_checked_ls(
first_block)
del iter
del inf
return len(positions_to_be_checked_ls)
