From: George George <george.githinji@...>
Date: 2009-10-16T13:25:52+09:00
Subject: Re: matching against a zillion patterns

Robert Klemme wrote:
> 2009/10/15 George George <george.githinji@gmail.com>:
>>
>>
>> depending on which pattern the string contains.
> Looks like this is tricky to become right.  There seem to be some
> people around that do bioinformatics, maybe some of them do have a
> solution already.
> 
> Things you could do off the top of my head: since it seems your
> patterns are only strings you could optimize the pattern by building a
> trie from it and then creating a RX from that.  I've done this before
> (see below).  Advantage is that your regular expression gets smaller
> and matching becomes more efficient (because of eliminated
> backtracking for NDA based RX engines).

Thank you so much Robert! I have tried your approach and compressed the 
patterns to a single regular expression. (removed backtracking). Seems 
like a parser solution approach may also help. e.g. the one that Josh 
had illustrated

Thomas, I have looked at bioruby and I regulary use it and relatively 
know it quite well, but we dont have  a solution that i am aware of. I 
might repost this on the bioruby list, together with a reference to this 
thread.
Thank you so much.

-- 
Posted via http://www.ruby-forum.com/.