Comments on Mining Drug Space: OpenAstexViewer - an open structural biology viewe...

Zsolt, thanks for the detailed update. It will tak...

2008-06-25T19:29:00.000+02:00

Zsolt, thanks for the detailed update. It will take me a while digging through it ...

Joerg,The training set we used for LASSO is also u...

2008-06-25T04:51:00.000+02:00

Joerg,

The training set we used for LASSO is also up on ChemSpider, for ACE it is at:

http://www.chemspider.com/docs/lasso/ACE.pdf

Seems like your test case does not match any of them exactly, but there are some similar ligands in the set. We have published test results for 40 families (the DUD set) recently, test sets are all disjunct from training sets:
JCAMD, vol 22, Num 6-7, June/July 2008, pp 479-487, doi: 10.1007/s10822-007-9164-5

We have included some docking result comparisons with other tools in our eHiTS papers:
J.MGM Vol 26, 1, July 2007, pp 198-212; doi:10.1016/j.jmgm.2006.06.002
Current Protein and Peptide Science, 2006, vol 7, pages: 421-435, Number 5, October 2006.

Unfortunately, we were not included in third party published evaluations, but asked some authors to perform the same tests with eHiTS after publications, e.g. Merck's comparative enrichment study by Wendy Cornell's group was repeated with eHiTS, you can see the results here:
http://www.simbiosys.ca/ehits/ehits_enrichment.html

Similarly, a published comparative validation study was repeated with the same data by one of our academic users Dr. Fedor Zhuravlev from Denmark:
http://www.simbiosys.ca/ehits/ehits_validation.html

There is also a list of publications by various users of eHiTS here:
http://www.simbiosys.ca/science/user_pubs/index.html

Zsolt

Zsolt, actually I am not too surprised that the AC...

2008-06-24T22:37:00.000+02:00

Zsolt, actually I am not too surprised that the ACE inhibitor was found, since I guess this one might have been part of the training set, was it?

Can you comment on this and the performance on molecules not being part of the training set?

Furthermore, are there any validation and performance comparisons of eHITS against other docking tools published, e.g. against GLIDE or MOE?

I am glad to see you are using the LASSO results o...

2008-06-20T17:13:00.000+02:00

I am glad to see you are using the LASSO results on ChemSpider and I am even happier that the activity estimate was correct. In general, LASSO is really meant to be a pre-filter in virtual screening, its task is to eliminate structures that do not have any chance of binding to the target. The remaining candidates should be fruther evaluated by an accurate docking tool, like eHiTS:
http://www.simbiosys.ca/ehits/

Zsolt Zsoldos (akak ZZ)

Joerg, Nice to see your comment about LASSO. A tec...

2008-06-20T04:56:00.000+02:00

Joerg, Nice to see your comment about LASSO. A technical note is in development right now that I believe will be quite impressive regarding the applications of LASSO to the identification of estrogen binders using ChemSpider.