Among the major features of 2.0 are spedup of up to 30% on the PC platform, completely re-worked static data files for improved chemistry, compatibility of the major modules -PHARM, -C, and FlexE, greatly enhanced FlexX-PHARM, a reworked Python-interface, SMARTS-support and the ability to read now all major file formats (MOL2, SDF, SLN, SMILES, CSLN, and PDB).

However, this is only the top-level functionality. A lot of improvement has been done behind the scenes to make FlexX even more reliable and robust. You'll find the detailed list of features here.

ToPNet release 1.3 offers the following new features:

	New Gene Expression GUI: A completely re-designed GUI for loading gene expression data now makes it very easy to integrate protein or gene networks with your expression measurements.
	GO-Annotation: Extracted sub-networks can be analyzed according to GO-terms. You get a selection of related GO-terms that are most specific for your network. This may be useful, e.g. for relating biological functions with selected Significant Areas to see which biological processes or molecular functions are affected by your expression experiments.
	Pathway Queries: This new feature lets you search for pathways that are interesting with respect to certain criteria. For example, you may look for 'druggable' targets like kinases that should phosphorylate a transcription factor regulating genes with a significant expression change in a certain experiment. To allow for such complex queries, ToPNet contains an XML-based query language that is used to formulate pathway templates by posing restrictions on network nodes and their connections.

Permute generates a small combinatorial library for each ligand (containing e.g. the different protonation states) and these libraries are docked using the FlexX^C-technology. Therefore Permute preserves the speed and comfort of single ligand docking.

The frequent issue of either generating multiple copies of the compound with the different protonation states, or a priori selection of a particular protonation state is a past issue with Permute. Last year we first published results on the basis of this new module at the 3^rd Joint Sheffield Conference on Chemoinformatics.

In this issue we take a closer look at the deepsite option. It is a well know problem that FlexX has a tendency to place ligands at the rim of the pocket so that oftentimes the binding pocket is not completely occupied. The user can actually do a lot to prevent such artifacts of the algorithm. A couple of tips and tricks to mitigate this problem at least to some extend can be found on the tips & tricks page.

If you have any questions or know of any tips and tricks yourself which you would like to share with the FlexX user community, we would appreciate your input at flexx-info@biosolveit.de.

NovoBench
BioSolveIT and Partners (Molecular Networks, ZBH, CCC, Eli Lilly, 4SC, and Altana Pharma) have started a joint project with the aim of an integrated suite for fragment based 'de novo' design of lead compounds. The 'integration' of the software extends from a 'synthesisability' module to the disappearance of fuss with different data formats. The project with a volume of € 2.000.000 won a three-year grant by the German Federal Ministry of Education and Research (BmBF).

ACS National Meeting in San Diego
BioSolveIT will be at this year's ACS National Meeting with two oral contributions

1. Mining docking space (COMP 310)
   Christian Lemmen et al.
   The talk will report the idea of our docking database (DDB) which is currently being interfaced to a generic toolbox for machine learning techniques. The graphical user interface makes the extraction of important information from hundreds of thousands of docked compounds a lot easier.
2. Boosting descriptors for similarity searches: feature trees trained by machine learning (CINF 83)
   Marcus Gastreich et al.
   This talk focuses on an application of the feature trees descriptor (FTrees) to large, real life datasets. The combination of FTrees with machine learning techniques leads to improved enrichments. The design of both the descriptor and the methodology deliver an information on what the important parts of actives are.

Combination of a Naive Bayes Classifier with Consensus Scoring Improves Enrichment of High-Throughput Docking Results
A.E. Klon, M. Glick, and J.W. Davies
J. Med. Chem.; (Letter); 2004; 47(18); 4356-4359
details here

Virtual Screening: finding needles in a haystack on a shoestring
D.E. Clark, N.V. Harris, A.G. Roach, and A.D. Baxter
Drug Discovery World, 2:37-41, 2004
details here