If you're going to worry about with and without decimal places (++ for that), try:

#!/usr/bin/perl

use strict;
use warnings;

use Regexp::Common;

while (<DATA>) {
    s|($RE{num}{decimal})%|$1 * 0.01|eg;
    print;
}

__DATA__
ENSP00000233379 1058 30 1206 1298 96.1% 13 + 96524483 96533474 8992
ENSP00000233379 1058 30 1206 1298 96% 13 + 96524483 96533474 8992
[download]

Ok, so you have to go install Regexp::Common. Getting this right every time is well worth that minor expense, IMO. Otherwise, if somehow the data gets corrupted so there is, say, "S%" in there somewhere, your code will generate an error. To take care of "S6%", just add a \b:

    s|\b($RE{num}{decimal})%|$1 * 0.01|eg;
[download]

It will leave the corruption alone this way. Detecting the corruption is as simple as checking if there are remaining %'s after the substitution:

    if (/%/) {
       print STDERR "Input has been corrupted!";
       # die ?
    }
[download]

Ok, maybe I'm going beyond the original post's scope here... ;-)