Comments on: Reviewing the CDK VFLIB patch

By: Constance21Thomas

Constance21Thomas — Mon, 05 Jul 2010 13:56:18 +0000

Make your life more easy take the loan and everything you want.

By: Christoph Steinbeck

Christoph Steinbeck — Fri, 26 Jun 2009 12:24:40 +0000

Hi Asad, thanks a lot for your comment and offer. We are currently sorting out with Rich a few glitches in his code. Please look at http://depth-first.com/articles/2009/06/16/if-the-wheel-doesnt-work-reinvent-it and the discussion therein. The patches resulting from this might be interesting for you. I’ll point Mark Rijnbeek in my group to your offer. Cheers, Chris

By: Syed Asad Rahman

Syed Asad Rahman — Thu, 25 Jun 2009 23:14:29 +0000

Hi Chris,

I have implemented VFLib code from Rich (avoided reinventing the wheel) into SMSD for Substructure search. It works pretty well.

The code is very much in object oriented format and I have made necessary changes in the code to work with CDK.

Would be happy to share the code if someone is interested.

Asad

By: Christoph Steinbeck

Christoph Steinbeck — Tue, 16 Jun 2009 17:43:39 +0000

Rich, thanks so much for the great comment. We’ll definitely have a look at it. I will point it our to Mark, my OrChem developer. He uses Rajarshi’s VFLib code extensively but would certainly welcome some maintainable code. Cheers, Chris

By: Rich Apodaca

Rich Apodaca — Tue, 16 Jun 2009 16:52:12 +0000

“I checked it out and looked at the code, which looked horrible because it was a 1:1 translation of a horrible looking C code. Clearly, a decent naming of the variables would greatly improve the code but I remember a statement that the translator himself could not make sense out of this, so the original author is to blame . I do not get the impression that this problem can be rectified quickly. In fact, it took Mark a few days to debug this code by adding a rich collection of debug messages. I’m not sure that this is how it should be. The code is essentially unreadable.”

Chris, my point exactly:

http://depth-first.com/articles/2008/11/13/one-of-these-things-is-not-like-the-other

Rajarshi and I both made attempts to port the C code. The C code might make sense to a C programmer, but Java is really a completely different language with a totally different paradigm. Rajarshi made it to the end, but I gave up in despair. There was no way I wanted to be fighting with code like that for the next few years.

The reason I gave up on porting: unreadable code is unmaintainable and non-extendable code. It promotes bitrot, no matter how well-tested it is.

Then I discovered that the documentation on the VF algorithm is surprisingly easy to read. The algorithm lent itself very nicely to a simple, object-oriented implementation:

http://amalfi.dis.unina.it/graph/db/papers/vf-algorithm.pdf

This is why I created a Java-centric implementation of the VF algorithm from scratch for MX:

http://depth-first.com/articles/2008/11/13/one-of-these-things-is-not-like-the-other

Implementation:

http://github.com/rapodaca/mx/tree/00b9b4e8ee0627a8a4a965cebe510ecb50a5d6cd/src/com/metamolecular/mx/map

Main unit tests are here:

http://github.com/rapodaca/mx/blob/00b9b4e8ee0627a8a4a965cebe510ecb50a5d6cd/src/com/metamolecular/mx/test/DefaultQueryMapperTest.java

High-level documentation of the code is available here:

http://depth-first.com/articles/2008/11/17/substructure-search-from-scratch-in-java-part-1-the-atom-mapping-problem

It’s MIT licensed, so feel free to port if you’d like.