New algorithms and software for analyzing and classifying evolutionary and biomedical data

点击翻译按钮获取中文摘要

My research proposal involves five major components related to the development of new algorithms and software for analyzing and classifying evolutionary and biomedical data. First, we will continue to investigate the phenomenon of reticulate evolution in the context of horizontal gene transfer (HGT). We propose to design original and effective maximum likelihood algorithms for inferring and validating statistically both complete and partial HGT events as well as for determining types (i.e., whether the transferred gene is additive, replacing or recombining) and units (i.e., whether gene transfer involves gene fragments, whole genes or entire operons) of gene transfers. The proposed algorithms will be used to estimate the impact of HGT on the resistance of bacteria to antibiotics - a topic of particular interest to the Canadian healthcare system and pharmaceutical industry. Furthermore, we will develop and maintain an up-to-date database dedicated to gene transfer networks of antibiotic resistance genes. Second, we will design a novel bioinformatics framework for estimating and validating the rates of complete and partial HGT among prokaryotes at different phylogenetic and ecological levels. It will allow researchers to determine the proportion of mosaic genes in prokaryotic genomes, to identify prokaryotic families and habitats being the major donors and recipients of genetic material, and to assess the ages of the detected HGT events. Third, we will propose a novel maximum likelihood method for identifying diploid hybridization events, including statistical validation of the detected hybrids and their parents by bootstrap analysis. This method will be extended to determine whether the relationship among the given species should be represented by a phylogenetic tree or by a hybridization network. Such methods will be of significant interest to a large community of plant and fish biologists. Fourth, we will design novel algorithms and a new statistical test for analyzing and correcting experimental high-throughput screening (HTS) data. This test will identify the type of systematic bias affecting a given HTS assay (i.e., additive or multiplicative bias). The new algorithms will be used to detect and eliminate multiplicative type of systematic bias in experimental HTS. Moreover, a novel data processing protocol for optimizing hit selection process will be introduced. The proposed methods will allow researchers to minimize the impact of systematic error in experimental HTS and thus improve the selection of potential drug candidates. Finally, we will develop open-source software to allow academic and industrial researchers throughout the world to carry out our new algorithms for detecting, validating and visualizing HGT and hybridization events, inferring and examining gene transfer networks of antibiotic resistant genes, and correcting and analyzing experimental HTS assays.