Tuesday, August 05, 2008

Open Access to calculated chemical properties of 12M substances

ChemStar (Computed Molecular Data) is an Open Access database of the calculated chemical properties of over 12M substances from PubChem. The process - which demands significant computing power - was done is a distributed fashion using JavaRMI and is described in the paper below. The properties are calculated using three different methods: JChem, JOELIB and MOE. The site allows the perusal of the properties, with a Marvin-based applet for viewing structures. The Java sourcecode for the distributed computing is available.


Karthikeyan, M., Krishnan, S., Pandey, A.K., Bender, A., Tropsha, A. (2008). Distributed Chemical Computing Using ChemStar: An Open Source Java Remote Method Invocation Architecture Applied to Large Scale Molecular Data from PubChem. Journal of Chemical Information and Modeling, 48(4), 691-703. DOI: 10.1021/ci700334f

2 comments:

ChemSpiderMan said...

I tried to get to ChemStar but it didn't seem to be up tonight. Looking forward to reviewing it. FYI we are presently adding the EPISuite predictions to the 21.5 million compounds on ChemSPider: http://www.chemspider.com/blog/adding-epa-episuite-properties-to-chemspider.html

Glen Newton said...

I have had periods where connectivity to ChemStar have been slow, but no complete failures.

I just tried it [10:23 Eastern time] and is up.