Cleavage knowledge extraction in HIV-1 protease using hidden Markov model

Inactive HIV is a poly protein precursor. This protein chain has to be cleaved at 9 specific positions to produce individual functional mature proteins responsible for making up a new active virus. Cleavage knowledge extraction in HIV protease will assist in designing effective inhibitors used in the treatment of AIDS. Although much progress has been made in sequencing the viral protease, little progress has been made in understanding the specificity. Several machine learning techniques have been used in understanding the specificity of HIV-1 protease with the highest prediction rate being 92%. In this paper the hidden Markov model is used for analyzing the specificity of HIV-1 protease. The objective is to learn the lock and key mechanism of the protease and protein precursor using the hidden Markov model (HMM) from a set of experimental observations. A good self consistency rate of 96% and recognition accuracy of 95.24% on unseen data is achieved. The HIV protease specificity to cleave between a phenylalanine and tyrosine or proline is also validated by our experiments indicating that the HMM is successful in learning the complex lock and key rule between protease and precursor protein. Used with other techniques, HMM can be used as an effective tool for designing new drugs.