Matching '=' and other non alphanumeric characters using regular expressions

[Anonymous Monk]

I am attempting to extract information from a text file across a number of lines.I can't find the correct regular expression syntax to match when it comes across = or () or '.' or any other non-alphanumeric characters. '.' only seems to match alphanumeric characters.Can anyone help me? ie:

#! /usr/bin/perl

use strict;

my $string = "LOCUS SOY4KDPP 545 bp  PLN 01-FEB-2000
DEFINITION  BIO5 gene complete cds EC=1.7.7.1.
ACCESSION   D17396
VERSION     D17396.1  GI:498167
KEYWORDS    protein; insulin-like protein;
            leginsulin.
//";

if ($string =~ /DEFINITION\s\s([\w\s]+)ACCESSION/){
    
    print "$1\n";
    
}

exit;

[pg]

When I looked at your string, I realized that it is actually a kind of key-value pair. My suggestion is to save the result returned from m// into a hash.

Make some small modification to the following code I gave, you can easily make it work for you:

use strict;

my $string = "DEFINITION  BIO5 gene complete cds EC=1.7.7.1.
ACCESSION   D17396
VERSION     D17396.1  GI:498167
KEYWORDS    protein; insulin-like protein;";

my %KVpairs = ($string =~ /(\w+)\s+(.*)/mg);#save the result right int
+o a hash
foreach (keys %KVpairs) {
   print "[$_] = $KVpairs{$_}\n";
}

[MarkM]

'.' will (normally) match all characters except for the newline character.

The regular expression that you are looking for is probably:

if ($string =~ /^DEFINITION[ \t]+([^\r\n]*)/mg) {


In english, this would be: A line in $string that begins with the literal string 'DEFINITION' followed by any amount of simple white space (' ' and '\t') followed by a string of characters that do not include '\r' or '\n'. This will grab all characters to the end of the line, but not the end-of-line character sequence itself.

I choose to use [^\r\n]* instead of .* as I regularly have to ensure that my code will function equally well under both UNIX and WIN32. Using [^\r\n]* instead of .* allows me to ensure that '\r' is not picked up at the end of $1.

[blahblahblah]

In addition to what the others have pointed out, I think you'll need to add an 's' to the end so that you can match across multiple lines, like this:

if ($string =~ /DEFINITION\s\s([\w\s]+)ACCESSION/s){
    print "$1\n";
}

Also, if you simply want to pick up everything between "DEFINITION " and "ACCESSION", another way to do it without worrying about knowing every possible character is like this:

$string =~ /DEFINITION\s\s(.*?)ACCESSION/s

[krujos]

you need to match the . and = characters. If there is a genral form for the where thoes occur you could plug them into your scripts. Without knowing that a smiple suggestion is to just add the = and periods to your regex.

if ($string =~ /DEFINITION\s\s([\w\s=\.]+)ACCESSION/){

Good luck. Josh

[Bukowski]

Hmm I smell a bioinformatics question :)

You're obviously trying to parse out information from GenBank records.

Have you come across Licoln Steins Creating a Bioinformatics Nation?

One of the points is that there is massive duplication of effort. Every Perl using biologist has written a GenBank parser!! I have, my colleagues have.. friends of colleagues have!

Run, don't walk, to BioPerl and save yourself the time and effort - the parsers you require are already there for you :)

Bukowski - aka Dan ([e-mail])
"Coffee for the mind, Pizza for the body, Sushi for the soul" -Userfriendly

[jjdraco]

\W matches a non-word character
\C matches a character


jjdraco
learning Perl one statement at a time.