ANNAVP, using neural networks to predict neutralization efficiency of antibodies against viral strains and to cluster strains by protein sequence

Abstract

Studying viral antibody neutralization data is a complex task and knowledge relating to the effectiveness of a particular antibody to particular strains of viruses cannot easily be extrapolated to other new, related strains. We have developed ANNAVP, a software that uses neural networks to model viral protein data. ANNAVP uses supervised or unsupervised learning and viral protein sequence data to form correlations between different strains and to predict the effectiveness of neutralizing agents against them.