Re^2: Is there a way to compare strings without using an array?

[Jeri]

in reply to Re: Is there a way to compare strings without using an array?
in thread Is there a way to compare strings without using an array?

Actually I'm not sure this will work because my string will look something like this..

648040620,637132715,649986572 etc.

Levenshtein could take into account some unwanted modifications, such as changing the entire 9 digit number by changing a single digit

I need the numbers not to be considered as change-able

[CountZero]

Wait ... you mean your proteins are not single letters? But actually a 9 digit number like 648040620? Then the regex solution as mentioned above will not work and I think you better put your data in arrays and use List::Compare to calculate the set intersection and such.

[Jeri]

I like the List::Compare. It looks good. I'm just worried about, because my array will be at least 5 million in length. Can you offer me any peace of mind?

[tj_thompson]

Abstract your comparison code out into a subroutine and give it a try. Should be quick simple code to write and you'll know exactly how well it performs. Will make changing the code or comparing it to other possible solutions simple as well.

[CountZero]

Here is some code which does what you want:

use Modern::Perl;
use List::Compare;

# Make two arrays with fake data
my @left  = map {'AA' . int(rand (5000000) + 1000000)} (1 ... 5000000)
+;
my @right = map {'AA' . int(rand (5000000) + 1000000)} (1 ... 5000000)
+;

my $start = time;
my $lc = List::Compare->new('-u', \@left, \@right);

{
    open my $OUT_INTER, '>', './intersection.txt';
    say $OUT_INTER $_ for $lc->get_intersection;
}
{
    open my $OUT_DIFF, '>', './difference.txt';
    say $OUT_DIFF $_ for $lc->get_symmetric_difference;
}
my $duration = time - $start;

say "Duration: $duration seconds.";

However, on my small laptop (only 2 GByte RAM and running a lot of other programs and services) I get an "Out of memory" error, so two lists of 5 million proteins is too big for it. YMMV if you have a more performant machine. It runs fine and fast for 500,000 proteins (15 seconds), 1,000,000 proteins (52 seconds) and 2,000,000 proteins (185 seconds).

[mrstlee]

I'm not 100% sure what you want to do here Jeri. The sample below assumes you have a single string composed of 9 character sequences. When the 'for' has run the keys of %uniq will be the unique 9 character sequences.

It would take a while and a lot of memory over 5 mill x 9 char sequences!

my $str = join '',(648040620,637132715,649986572,648040620 );

my $proteins_count = length ($str)/9;

my %uniq;

do { $uniq{$_}++ unless $uniq{$_} } 
 for unpack "(A9)$proteins_count" , $str;

print "@{[keys %uniq]}\n";

Prints ..
637132715 648040620 649986572

[Jeri]

I'm going to give it a shot today. I'll let you know what happens. Thanks.