Classification of HIV-1 Using Coalescent Theory

该项目的目的是开发一种自动分类HIV-1的方法。该方法的主要部分包括对给定分类进行评分的评分算法。该算法在给定分类条件下重构给定HIV-1序列的祖先重组图（ARG），并利用马尔可夫链蒙特卡罗方法找到具有最大概率的祖先重组图。最可能发生ARG的概率被解释为分类的分数。然后，根据分数迭代改进分类。在准备阶段，我们将改进jpHMM算法，这是我们团队开发的用于定位病毒基因组重组的工具，在运行时间和精度方面。最后，我们将系统地应用我们的方法重新注释美国洛斯阿拉莫斯国家实验室（Los Alamos National Laboratory， LANL）的HIV-1基因组数据库。分类算法的开发、评估和应用将与艾滋病病毒序列分析和进化领域的两位权威专家、ll的B. Korber博士和T. Leitner博士以及加拿大滑铁卢大学隐马尔可夫模型的知名专家D. Brown教授密切合作。

The aim of the project is to develop a method for automatically classifying HIV-1. The main part of this method consists in a scoring algorithm which rates a given classification. This algorithm reconstructs ancestral recombination graphs (ARG) of given HIV-1 sequences under restrictions determined by the given classification and finds an ARG with maximal probability by means of Markov Chain Monte Carlo methods. The probability of the most probable ARG is interpreted as a score for the classification. Afterwards, the classification is iteratively improved with respect to the score. In a preparing step, we will improve the jpHMM algorithm, a tool developed by our group to localize recombination in viral genomes, with respect to runtime and precision. Finally, we will systematically apply our method to re-annotate the HIV-1 genome database at Los Alamos National Laboratory (LANL). Development, evaluation and application of classification algorithm will be carried out in close collaboration with Dr. B. Korber and Dr. T. Leitner at LANL, two leading experts on HIV sequence analysis and evolution, and with Prof. D. Brown, University of Waterloo, Canada, a well-known expert on Hidden Markov Models.