Pred-hERG is based on statistically significant and externally predictive QSAR models of hERG blockage. Inferring new instructions from data is the core strength of machine learning. It also highlights the critical role of data: the more data available to train the algorithm, the more it learns.
Developed as a tool for identifying putative hERG blockers. The consensus models were generated averaging the predictions of individual models, achieving balanced accuracy, sensitivity, and specificity as high as 89%-90%.
The largest publicly available dataset for hERG liability was retrieved from the ChEMBL 23 database containing 16,932 associated bioactivity records of hERG K+ blockage for 8,531 unique organic compounds. The curated dataset used in our publication containing 5,984 compounds is available for download here
The probability maps allow the visualization of predicted fragment contribution. This method provides an easy interpretation of the predicted activity, allowing the user to easily propose structural modifications.
Predictions are fast and predictions for one chemical appear directly on the screen. Indeed, simpler models (e.g. linear instead of non-linear, or with fewer parameters) often run faster but are not always able to take into account the same exact properties of the data as more complex ones.
Our models also satisfy the guidelines of the Organisation for Economic Cooperation and Development (OECD) principles and the list of tests required by Registration, Evaluation, Authorisation and Restriction of Chemical substances (REACH) for successful toxicity assessment.
Click the right button on the whiteboard of the "Molecular Editor" and select "Paste MOL or SDF or SMILES"." SDF and MOL files are accepted.
Our Headquarter is in Brazil