An Interesting New Paper

Titled: “Distributed Chemical Computing Using ChemStar: An Open Source Java Remote Method Invocation Architecture Applied to Large Scale Molecular Data from PubChem.”

Authors: M. Karthikeyan, S. Krishnan, Anil, Kumar Pandey, Andreas Bender, and Alexander Tropsha.

Provisionally published in JCIM.

A long story short, they have published the first online open source distributed computing and data analysis architecture written in Java using the remote method invocation. The resource can be found at:

http://moltable.ncl.res.in/chemstar/index.jsp

The network consists of 25 windows based machines. Server side computing requires Java, Oracle, MySQL or MS Access. On the client side only Java is required.

They have demonstrated its ability by calculating descriptors on a dataset of over 11 million compounds using the JOELib and Marvin toolkits. At present they are able to calculate: pKa, log P, log D, polar surface area, charge distribution, polarizability, topology analysis, H-bond acceptor, H-bond donor properties, the prediction of protonation microspecies, Huckel analysis, refractivity and elemental analysis. They envisage extending these methods to in-silico-based virtual screening, fast docking, text mining or other related informatics applications.

I am also very interested in this work as their homepage (see link above) is written in Java Server Pages (JSP), a language I have started to write some open source web services in.