Pred-hERG

Fast hERG Liability Predictor

The Pred-hERG is a freely available web service for evaluation of cardiac toxicity via hERG inhibition. The predictions are made by predictive and extensively validated QSAR models. The Pred-hERG server employs many tools including flask, uWSGI, nginx, Python, and JavaScript. Towards more user-friendly interface, Pred-hERG provides an interactive web interface, including the JSME, a free molecule editor in JavaScript ,1 written on JavaScript and support the latest versions of all most popular browsers. Users will not need any Java or Flash plugins to use it in their browser. Pred-hERG is implemented on Ubuntu Server.

The new version of the web server  has a simpler intuitive user interface. The user needs to paste the SMILES string of the structure in the appropriate area and hit ‘Predict’ button. After the completion of the job, the user will receive the outcome with the predicted probability maps in the page.

Lightning fast

Pred-hERG is based on statistically significant and externally predictive QSAR models of hERG blockage. The models were built using the largest publicly available dataset of structurally diverse compounds including variety of drug classes. The models developed can be used by the research community and regulatory scientists for the rapid evaluation of cardiac toxicity liability via hERG inhibition in chemical inventories through Pred-hERG.  All curated datasets and developed models been made publicly available .

Pred-hERG v. 3.0

Pred-hERG is now on version 3.0. For a full list of changes, please see release notes.

Release Notes

How to cite

If this service was useful to you, please cite:

1. Braga, R.C.; Alves, V.M.; Silva, M.F.B.; Muratov, E.; Fourches, D.; Liao, L.M.; Tropsha, A.; Andrade, C.H. Pred-hERG: A novel web-accessible computational tool for predicting cardiac toxicity. Mol. Inf. 2015, just accepted.

2. Braga, R. C.; Alves, V. M.; Silva, M. F. B.; Muratov, E.; Fourches, D.; Tropsha, A.; Andrade, C. H. Tuning hERG out: Antitarget QSAR Models for Drug Development. Curr. Top. Med. Chem. 2014, 14, 1399–1415. 10.2174/1568026614666140506124442

Reference

  • Dataset

    The largest publicly available dataset for hERG liability was retrieved from the ChEMBL 21 database containing 16,932 associated bioactivity records for the hERG K+ channel. Also, we used inactive compoudns from the PubChem BioAssay 1511 and unique compunds from Li et al. paper. Since the previous version we have increased 2,535 bioactivity records for the hERG K+ channel. Once curated, this dataset contained 8,552 unique compounds, with more than 2.5k new compounds compared with the previous version (PredhERG 2.0).

  • QSAR models

    QSAR models were developed as virtual screening tools for revealing putative hERG blockers. Consensus models were generated averaging the predictions of individual models, achieving balanced accuracy, sensitivity, and specificity up to 0.89-0.90 with the coverage of 0.63-1.

  • Similarity maps

    This method allows visualizing how a fragment can contribute to the activity of the compound (positively or negatively) and its original code has been implemented in the server.

  • Study Case

    The WDI dataset (version 2010) with almost 53,965 chemical compounds, including all marketed drugs and compounds that entered clinical trials, was evaluated. Please read our paper for additional information.

Behind the Server 

flexqsar

Additional Information

Support Information

Here you find the complete  statistical results of generated QSAR models for the modeling set. For additional information, please read our paper.

Pred-hERG_Supplemental_materials

Curated dataset

5,984 unique compounds retrieved from ChEMBL. Additional information about source and curation procedures:

5984-dataset-binary-curated.sdf (SMILES)

Similarity Diversity

Dendrogram and heat map for the 5,984 compounds suggested a high level of structural dissimilarity.

Similarity Diversity

Terms of Confidentiality

The information sent to Pred-hERG is confidential. The web server safeguard and keep the structures sent in the strictest confidence at all times. We do not disclose or divulge any of the confidential information to any third party. We take all reasonably available measures to preserve the confidentiality of, and not to disclose, any information sent by users. The server do not save any data regarding the structures or IP of users. All data is automatically erased from the server as soon as the job is done and the report sent.

Financial Support

References:

(1)  Bienfait, B.; Ertl, P. JSME: A Free Molecule Editor in JavaScript. J. Cheminform. 2013, 5 (5), 1–6.

Karulin, B.; Kozhevnikov, M. J. Cheminform. 2011, 3, P3.

(2) Riniker, S.; Landrum, G. a. J. Cheminform. 2013, 5, 43.

(3) Li, Q.; Jørgensen, F. S.; Oprea, T.; Brunak, S.; Taboureau, O. Mol. Pharm. 2008, 5, 117–127.