A vectorized sequence-coupling model for predicting HIV protease cleavage sites in proteins.

What kind of peptide sequences can be cleaved by HIV protease, and what kind cannot be? This is a crucially important problem in designing effective inhibitors against HIV protease as potential drugs for AIDS therapy. To tackle this problem, a sequence-coupling and vectorized model is proposed for predicting the cleavability of oligopeptides by proteases with multiple and extended specificity subsites. In comparison with existing methods, the new method has proved to be an improvement in both the accuracy of the model and the rationality of the statistical treatment. Meanwhile, the Monte Carlo sampling procedure introduced here has also proved to be very useful in dealing with the situation when the experimental data are insufficiently sampled for complete statistics. Owing to its very high rate of correct prediction, it is expected that the new method can be a useful technique for helping to find effective inhibitors of HIV protease, which is one of the targets in designing potential drugs against AIDS. The principle of the new method can also be applied to analyzing the specificity of any multi-subsite enzyme.