Apart from what other people have said, i thought i should add that many of the things you want to do are fairly standard and there is a lot of help out there already - check out perl and bioinformatics as a start.

On bioperl you will propbably find there are parsers for your format, and once you have your sequences as a standard object, there are many methods for retreiving subsequences / mathcing etc...

If you files are really that huge, i would say that you probably need to process them one record at a time, or use a database - again there is help on bioperl for doing this, but i think you probably won't need to go that far.

Let us see what you have so far in terms of code, and an example of input and desired output too. HTH.