MDSC 408 multiple sequence alignment colourer

gistfile1.py

def main():
  lines = open('alignment.aln').readlines()
  lines = [line for line in lines if len(line.strip()) > 0]

  header = lines[0]
  del lines[0]

  num_lines = 3

  for k in range(0, len(lines), num_lines + 1):
    block_lines = lines[k:(k + num_lines)]
    classes = lines[k + num_lines]

    if len(block_lines) != num_lines:
      print block_lines
      raise Exception('Wrong number of lines')
    if not (len(block_lines[0]) == len(block_lines[1]) == len(block_lines[2])):
      raise Exception('Lines are not same length')

    colours = {
      'same_0_1': '#8fb9ff',
      'same_0_2': '#ff8f8f',
      'same_1_2': '#8dfba4',
      'same_0_1_2': '#f8ff2b'
    }

    coloured = []
    seq = []
    for i in range(len(block_lines)):
      line = block_lines[i].strip()
      #components = re.search(r'^([^\s]+[\s]+)([^\s+])([\s]+[\d]+$', line)
      block_lines[i] = block_lines[i].strip().split()[1]
      coloured.append('')

    for i in range(len(block_lines[0])):
      if aa_equal(block_lines[0][i], block_lines[1][i]) and aa_equal(block_lines[1][i], block_lines[2][i]):
        matching = (0, 1, 2)
      elif aa_equal(block_lines[0][i], block_lines[1][i]):
        matching = (0, 1)
      elif aa_equal(block_lines[0][i], block_lines[2][i]):
        matching = (0, 2)
      elif aa_equal(block_lines[1][i], block_lines[2][i]):
        matching = (1, 2)
      else:
        matching = ()

      colour = 'same_%s' % '_'.join([str(m) for m in matching])
      for j in range(len(coloured)):
        if j in matching:
          new_char = apply_colour(block_lines[j][i], colours[colour])
        else:
          new_char = block_lines[j][i]
        coloured[j] += new_char

    print '\n'.join(coloured)
    #print classes
    print ''

def apply_colour(char, colour):
  return '<span style="background-color: %s">%s</span>' % (colour, char)

def aa_equal(a, b):
  return a == b and a != '-'

print '<pre>'
main()
print '</pre>'