Update: Realisation dawned

The solution is:

#! perl -slw
use strict;

sub fact{
    my $n = shift;
    my $f = $n; $f *= $n while --$n;
    return $f;
}

sub nCr{
    my( $n, $r ) = @_;
    return fact( $n ) / ( fact( $r ) * fact( $n - $r ) );
}

sub NwithM {
    my $f = 4-1;
    my( $L, $M ) = @_;
    my $T = 1;
    for my $m ( 1 .. $M ) {
        $T += nCr( $L, $m ) * $f**$m;
    }
    return $T;
}


print "8 with $_ mismatch(es): ", NwithM( 8, $_ ) for 1 .. 4;
print "9 with $_ mismatch(es): ", NwithM( 9, $_ ) for 1 .. 4;
[download]

Produces:

C:\test\humanGenome>fuzzyDNAcalc.pl
8 with 1 mismatch(es): 25
8 with 2 mismatch(es): 277
8 with 3 mismatch(es): 1789
8 with 4 mismatch(es): 7459
9 with 1 mismatch(es): 28
9 with 2 mismatch(es): 352
9 with 3 mismatch(es): 2620
9 with 4 mismatch(es): 12826
[download]

All my searches turned up algorithms for generating these sequences, but I haven't found, or been able to derive a formula for calculating them.

Given a string of (a|c|g|t)x8; and allowing for up to 2 mismatched characters, how many possible matches are there.

There are 4^8 = 65536 possible 8-base sequences.

By experiment I know that allowing for

• M=2 mismatches, there are 277 possible matches for any given 8-base string.
• M=3 mismatches, there are 1789 possible matches for any given 8-base string.
• M-4 mismatches, there are 7459 possible matches for any given 8-base string.

I know I am probably missing the obvious, but I'm not seeing the formula?

If it helps, for L=9, total sequences is 4^9 = 262144; and the number of matches for various values of M is:

• L=9, M=2, T=352
• L=9, M=3, T=2620
• L=9, M=4, T=12826

Do you know, or are you able to derive the formula in terms of T = f( L, M )?

>Thanks.