I suspect, based on previous similar questions of this type, that what the op has is:

1. A bunch of long sequences; possibly 1000s or 100,000s bytes/codons/other long.
2. A bunch of shorter sequences perhaps 10 chars, perhaps 9 or 10 chars, long.

And the process he's trying to code is:

• For each of the long sequences...
• For each of the short sequences...
• Scan the longer sequence looking for sites where the shorter sequence 'matches' and record those positions.

The complication is that the matching is 'fuzzy' within his set of constraints:

• such that a 9 character subsection of the larger sequence may be considered a match to a 10-character short sequence, if (for example) the removal of any (exactly one) character from the 10 character sequence allows it to match the 9 character subsection of the longer sequence.
• Or, a 10-character short sequence may be considered a match for a 10-character subsection of the longer sequence if they differ in exactly (only) one position.

But, that's just supposition until he answers somebody's questions!