Computational analysis of HIV-1 evolution and drug resistance.

摘要

One major problem in the treatment of the acquired immune deficiency syndrome (AIDS) is the rapid development of drug resistance in its causative agent---human immunodeficiency virus (HIV). In order to better understand HIV drug resistance, I took advantage of the large amount of viral sequence data and the power of computational methods to study the evolution of two main drug targets in HIV, the protease and the reverse transcriptase in the pol gene. First, I globally distinguished and compared the effects of two confounding factors on HIV evolution: selection pressure on proteins and phylogeny. I systematically separated the covariation induced by selective interactions between amino acids from background linkage disequilibrium, i.e. phylogenetic effects, using synonymous vs. amino acid mutations. This study reveals that while there is a detectable level of background linkage disequilibrium, a large portion of covariation between amino acid mutations in the HIV pol gene results from drug-induced selection. Second, I studied in detail the evolutionary pathways of drug resistance in patients. I used within-patient viral sequence polymorphism, i.e. sequencing mixtures, to infer the temporal order of mutations. This prediction was validated by an independent longitudinal study, suggesting that the sequencing-mixture approach provides an economic alternative to conventional approaches, which require patient tracking. Finally, I investigated RNA secondary structures in HIV-drug target genes. RNA secondary structures have been discovered in different parts of the HIV genome; these structures have been found to play important roles in the viral life cycle. I applied a series of computational approaches to detect RNA secondary structures in the pol gene. Thermodynamic RNA folding predictions, synonymous variability analysis and covariance analysis each independently revealed strong evidence of a novel RNA secondary structure at the junction of the protease and the reverse transcriptase. The predicted structure was later validated in a biochemical probing assay. Discovery of this novel secondary structure suggests many directions for further functional investigations, which could be relevant to the development of more effective HIV/AIDS therapies.