Anti-HBV Predictor web server is designed to rapidly predict potent anti-HBV compounds with high accuracy. With more than 250 million chronically infected individuals worldwide, hepatitis B virus (HBV) remains a major public health threat. Current treatments often fail to achieve a functional cure, and the emergence of drug resistance continues to complicate the situation. One of the key challenges in HBV drug development is the ability to quickly identify potent anti-HBV compounds from the vast chemical space of billions of molecules. To address this challenge, we have developed a publicly accessible web server, Anti-HBV Predictor, which integrates high-throughput screening data to predict the anti-HBV activity of any user-supplied compound. The server provides predictions within seconds, with an accuracy exceeding 96% and an F1 score above 0.95. This enables large-scale virtual screening campaigns to discover novel anti-HBV agents and repurpose existing drugs. Anti-HBV Predictor is freely available to all researchers, supporting both single-molecule queries and batch screening of millions of compounds.

The Anti-HBV Predictor web server is user-friendly and straightforward. To predict new anti-HBV compounds, simply enter the compounds of interest in the input box, then click the "Predict" button and wait for the output results. Each line should contain the compound's SMILES notation and a one-word name, separated by a tab or space.

Anti-HBV Predictor requires approximately 20 seconds to load the necessary background data and trained models for prediction. Therefore, predicting the activity of 1-10 compounds takes less than 30 seconds, while screening up to 100 compounds takes about 2 minutes.

Input example:
[C@@H](CC(=O)N1CCN2C(C1)NNC2C(F)(F)F)CC3CC(F)C(F)CC3F Sitagliptin
OC1CCCC(\\C=C/2\\SC(=S)N(CC3OCCC3)C2=O)C1O CHEMBL1958457
OC(COC1(CCCCC1)C#C)CN2CCN(CC2)C3CCCC(C3)C(F)(F)F Terciprazine

Out example:
Click to view example output results