Beefy Boxes and Bandwidth Generously Provided by pair Networks
P is for Practical

Re: Spltting Genbank File

by citromatik (Curate)
on May 29, 2009 at 07:38 UTC ( #766822=note: print w/ replies, xml ) Need Help??

in reply to Spltting Genbank File

A Genbank file is made of a series of Genbank records separated by a line consisting of "//"

You are reading your file line by line:  while ($line = <$IN>) and splitting each line in different lines (??) my @line = split(/\///,$line). Obviously, this is not what you want.

An easy way to solve this is to read the file record by record by assigning the value "//" to $/ (see perlvar) and then process the record

Another possibility is to read the file line by line, and update the $accession, $biotype and $sequence variables, accordingly. But you can be in a problem if a record doesn't have one of them

Update: Following the first approach:

use strict; use warnings; $/ = "//"; my $genfile = "c:\"; open my $ifh, "<", $genfile or die "cannot open $genfile: $!\n"; while (my $chunk = <$ifh>){ last if eof $ifh; my ($accession) = $chunk =~ /LOCUS\s*([A-Z]*\d+)/; my ($biotype) = $chunk =~ /BIOTYPE\s*([A-Z])/; my ($sequence) = $chunk =~ /ORIGIN\s*(.*)$/s; $sequence =~ s/\s|\d//g; my $outfile = "${accession}_${biotype}"; open my $ofh, '>' $outfile or die "cannot open $outfile: $!\n"; print $ofh, ">$accession\n$sequence"; close $ofh; }

Additional comments:

  • You don't need to lc your input
  • I've never seen a "biotype" field in a Genbank file (I can't find it in the specs either), but I may be wrong, so I conserved the code that tries to capture that field, instead you can see the molecule type in the LOCUS field (DNA|PROT)
  • A typical "locus name" is made of letters (2 or 3) and numbers, you only test for 8 letters
  • The last record of the file is always empty (because the file ends with a "//") line, that is why I use last if eof $ifh;
  • Don't forget to close each output file when it is not needed, a Genbank file could be very large and you can run out of open filehandles, they are a finite resource on every operating system
  • The "ORIGIN" sub-record is multi-line, so you should add the /s modifier to your regexp (see perlre)
  • If you want to scape the "//" in a regular expression, you should escape both slashes: /\/\//
  • $sequence =~ s/\s//g; $sequence =~ s/\d//g; can be written $sequence =~ s/\s|\d//g;


Comment on Re: Spltting Genbank File
Select or Download Code

Log In?

What's my password?
Create A New User
Node Status?
node history
Node Type: note [id://766822]
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others musing on the Monastery: (15)
As of 2014-07-24 20:27 GMT
Find Nodes?
    Voting Booth?

    My favorite superfluous repetitious redundant duplicative phrase is:

    Results (167 votes), past polls