of the hERG inhibition (hERG): The human Ether-a-gogo Related Gene (hERG)
potassium channel is one of the major critical factors associated with QT
interval prolongation and development of arrhythmia called Torsades de Pointes.
Therefore it has become a growing concern of both regulatory agencies and
pharmaceutical industries who invest substantial effort in the assessment of
drugs cardiotoxicity. The prediction server calculates the hERG
inhibition based on the chemical structure information of the molecules. For the
prediction a Support Vector Machine (SVM) model trained on a 495 data set, and tested on two different external validation sets, is used. The model is built
on GRIND descriptors. The chemical structures can be fed to
the server in SMILES or sdf format, or sketched in
the JME applet. The structures are transformed to 3D sdf-format with the Molconvert
program from ChemAxon. Then molecules are washes, protonated and
minimized in vaccum, using MOE. The descriptors (using the DRY, Osp2, NH and TIP probes respectively) are generated with the Almond program from Molecular Discovery and fed to the WEKA-predictor.
The polynomial of degree 2 is set up with the SMOreg classifiers and non normalization in WEKA.
The biological activity associated to each compound is transformed to a binary system. The value "1"
define a hERG blocker compound and "0" a hERG non-blocker compound according to a threshold of 40 micromolar.
estimate for the herg prediction: For each tested compound a score and a confidence value is associated to. The score is between "-1" and
"1". A low score suggest a hERG non blocker and a high score a compound with a high chance to block hERG.
A simple confidence index is also assigned to the compound, for those who want a fast assessment.
"-" means predicted as non blocker, "low" for weak blocker and "high" for