1 #!/usr/bin/python
   2 # The line above is necessary to make program a valid UNIX script
   3 # Also necessary is format of newlines. If you use IDLE, it is probably
   4 # ok automatically. If you use FAR, you can convert newlines to UNIX
   5 # style either within FAR editor, or by command: sed -i 's/\r//g' file.py
   6 """ Simple FASTA manipulating example: find sequence by name in FASTA-file
   7 """
   8 import sys # useful in many places in file
   9 
  10 class Fasta_Sequence(object):
  11         """ Sequence retrieved from FASTA file """
  12 
  13         def __init__(self, string):
  14                 r""" Create object from chunk of FASTA file; leading ">" optional
  15 
  16                 Example:
  17                 >>> seq = Fasta_Sequence(">SEQ1\nMTEITAA\nLVSVK\nAAEVAAQL")
  18                 >>> print seq.name
  19                 SEQ1
  20                 >>> print seq.headers
  21                 None
  22                 >>> print seq.sequence
  23                 MTEITAALVSVKAAEVAAQL
  24                 """
  25                 # Initialize all fields in case 
  26                 self.name = None
  27                 self.id = None
  28                 self.headers = None
  29                 self.sequence = None
  30 
  31                 # Parse first line
  32                 first_line = string.split("\n", 1)[0]
  33                 if len(first_line.strip()):
  34                         self.name = first_line.split(" ", 1)[0].lstrip(">")
  35                         # By definition only sequence name may be present
  36                         if len(first_line.split(" ", 2)) == 3:
  37                                 self.id = first_line.split(" ", 2)[1]
  38                                 self.headers = first_line.split(" ", 2)[2].split(";")
  39 
  40                 # Parse sequence
  41                 self.sequence = "".join(string.split("\n")[1:])
  42 
  43         def write(self, file):
  44                 r""" Write sequence to a file
  45                 
  46                 Example:
  47                 >>> seq = Fasta_Sequence(">SEQ1\nMTEITAA\nLVSVK\nAAEVAAQL")
  48                 >>> seq.write(sys.stdout)
  49                 >SEQ1
  50                 MTEITAALVSVKAAEVAAQL
  51                 """
  52                 # Write headers
  53                 first_line = ">"
  54                 if self.name: # True only if self.name is not None or empty
  55                         first_line += self.name
  56                         if self.id:
  57                                 first_line += " " + self.id
  58                         if self.headers:
  59                                 first_line += ";".join(self.headers)
  60                 file.write(first_line + "\n")
  61 
  62                 # Write sequence, 80 chars per line
  63                 sequence = self.sequence[:]
  64                 while len(sequence) > 0:
  65                         file.write(sequence[:80] + "\n")
  66                         sequence = sequence[80:]
  67 
  68 def main():
  69         """ Check options, act accordingly """
  70         if options.seq_name == None:
  71                 print >>sys.stderr, "Option -s is mandatory!"
  72                 # This funny form of 'print' is quite like sys.stderr.write(...)
  73                 sys.exit(1)
  74 
  75         if options.fasta_file:
  76                 input = open(options.fasta_file)
  77         else:
  78                 input = sys.stdin
  79 
  80         if options.out_file:
  81                 output = open(options.out_file, "w")
  82         else:
  83                 output = sys.stdout
  84 
  85         for chunk in input.read().split("\n>"):
  86                 seq = Fasta_Sequence(chunk)
  87                 if seq.name == options.seq_name:
  88                         seq.write(output)
  89 
  90 if __name__ == "__main__":
  91         import optparse
  92 
  93         parser = optparse.OptionParser()
  94         parser.add_option("-f", "--fasta-file", help="Input FASTA-file name")
  95         parser.add_option("-o", "--out-file", help="Output FASTA-file name")
  96         parser.add_option("-s", "--seq-name",
  97                 help="Sequence name to search (mandatory)")
  98         parser.add_option("-t", "--test",
  99                 action="store_true", help="Run self-tests")
 100         parser.add_option("-v", "--verbose",
 101                 action="store_true", help="Verbose self-tests")
 102         options, args = parser.parse_args()
 103 
 104         if options.test or options.verbose:
 105                 import doctest
 106                 doctest.testmod()
 107                 sys.exit()
 108 
 109         main ()