Fuzzy Cognitive Maps for Modelling, Predicting and Interpreting HIV Drug Resistance

The high mutability of Human Immunodeficiency Virus (HIV) leads to serious problems on designing efficient antiviral drugs. In fact, in last years the study of drug resistance prediction for HIV mutations has become an open problem for researchers. Several machine learning techniques have been proposed for modelling this sequence classification problem, but most of them are difficult to interpret. This paper presents a modelling of the protease protein as a dynamic system through Fuzzy Cognitive Maps, using the amino acid contact energies for the sequence description. In addition, a Particle Swarm Optimization based learning scheme called PSO-RSVN is used to estimate the causal weight matrix that characterize these structures. Finally, a study with statistical techniques for knowledge discovery is conducted, for determining patterns in the causal influences of each sequence position on the resistance to five well-known inhibitor drugs.