eHiTS ®: Electronic High Throughput Screening
What's new in eHiTS ? Screening on a desktop supercomputer

eHiTS Lightning: is a newest & greatest version of eHiTS on the Cell/B.E. processor. Tremendous acceleration has been reached at no compromise to the accuracy. 

IBM's: white-paper, product brochure joint press release

summary page, white-paper , blog1, blog2

eHiTS Lightning (PS3 version) / eHiTS v9.1 (PC version) is released: see press-release. New features are documented in our Release Notes page. New results and how to obtain them are documented in our technical notes pages. Please contact us for more information.

eHiTS Lightning 2009 is the latest officially released version. It is more accurate, faster and simpler to use than ever before!
 Download product flyer

New Scoring Function

eHiTS has a novel scoring function that takes advantage of temperature factor information provided in PDB files to give a more complete picture of interactions.  All atoms in a PDB file have a temperature factor (B) associated with them.  This temperature factor indicates the how much the atom varies from the mean position.  Some atoms positions are very precisely defined while others vary greatly, this has a very strong influence on the weight that should be assigned to the position.  The novel approach in eHiTS uses the probability of the atom position during the statistic collection to create a statistically derived empirical scoring function.  The new scoring function unifies many of the components of old scoring functions and has provided a scoring function that is smoother and more accurately represents problem at hand.

High Accuracy

To see the effect of the new scoring function many validation studies were performed. Some were comparative studies to other previously published docking results, another study was designed by the developers of eHiTS, and that was performed on a set of 1586 pdb codes, i.e. protein ligand complex files taken from the PDB database.  These codes were selected as a combination of the GOLD validation set, the PDB-bind 2004 set and augmented with some more codes from the PDB, filtered  with the Lipinski rules for the ligands. In 100% of the cases, eHiTS 6.1 was able to reproduce the X-ray crystal pose within 2.5 Angstrom RMSD for the top ranking (scoring) pose found and the average top-ranking RMSD for the entire set was 1.1 Angstrom. For the closest pose, the RMSD was under 2.0 Angstrom in 100% of the cases and the average for the set was 0.73 Angstrom. This high level of accuracy has never been seen before in any docking program !   For more details of these validation studies please click here., or see our technical notes

eHiTS LASSO for pre-screening large databases

eHiTS now has a very efficient and accurate pre-screening tool for virtual high throughput screening of large databases.  The new eHiTS LASSO is able to screen 1 million ligands / minute, making it very suitable to pre-screening large databases.  The LASSO accuracy have been verified for four different families: cox-2, thrombin, estrogen, and gelantinase. For each set we picked 20 actives and 200 decoys for training and the total number of actives (used for test runs) in each set was 128, 67, 55, and 43 respectively.  There were one thousand decoys in the test runs as well. Our results show that for all sets the first 10 percent ligands in the filter ranked list, contain more than 60 percent of the actives and the first 20 percent has more than about 80 percent of actives (for thrombin this was 74.6 percent).

Improved Database Handling

The eHiTS docking algorithm allows it to re-use docking results of rigid fragments found in one ligand, when docking subsequent ligands.  This can greatly reduce the time required to perform a large database screen. This is due to the fact that many rigid fragments are repeated in a ligand database, thus by storing the docking information and simply looking it up when needed, the time required to perform a docking calculation is greatly reduced.

To test the effect of the database on a virtual ligand screen, 5000 ligands were screened against an estrogen receptor (pdb code 1ERR). The 5000 ligands were randomly selected from a set of drug-like ligands obtained from the ZINC database of ligands.  The plot below shows the speed-up found as the database is filled up, i.e. more ligands have been screened.

Here we can see that the docking speed / ligand goes from 2.4 mins / ligand (with no DB) down to under a minute per ligand once most of the "common" fragments have been added to the database.

Support for distribution within a cluster environment

eHiTS has support for the LSF, SGE, and PBS queuing system.  This allows users to take advantage of the computational power available on clusters requiring the use of this queuing system.  eHiTS has always been able to run over a cluster environment, however this new feature allows users to conform to system administrators requirements.  In addition, SimBioSys is able to modify eHiTS to comply with any queuing system you may have on your cluster, please contact us for more information.

Other Advantages of eHiTS

Automatic and Fixed Protonation state handling

eHiTS was the first docking program that evaluated all possible protonation states of the ligand and receptor in a single run, that we know of.  Please see FAQ for more information on protonation handling in eHiTS.  , or see the Automatic portonation state handling technical note

Starting from version eHiTS 2009 protonation states can be fixed, allowing expert users more control over this important information.

Our Scoring Function can be trained by the user

eHiTS contains training scripts that allow the user to train the scoring function with data they may have available.  Both validation training (using co-crystallized complexes) and enrichment training (using known binders and known decoys) are available in eHiTS.  Training the scoring function can dramatically improve the ability of eHiTS to identify the correct binding pose or to identify active ligands from decoy ligands.