Toward Structural and Functional Predictions from Biological Sequences

Biological sequences, including DNA and protein sequences, are believed to encode sufficient information to determine the structure and function of biological molecules, which in turn decide the phenotypic traits of animals. Deciphering the biological sequences is an important and multiscale problem that connecting the information flow from genotypes to phenotypes. Current advances in next-generation sequence technology provided tons of sequencing data, demanding innovations in computational algorithm for better interpretation. I developed computational methodologies to understand the biological sequences in various levels. In the primary sequence level, I analyzed the evolutionary information encoded in protein families and predicted the function (and active sites) of the proteins. To aid my sequence analysis, I developed a set of computational methodologies and deployed them as public web-servers. In the protein structure level, I studied the plasticity of the 3D structures, as well as demonstrated its effect on the uncertainty of computational scoring algorithms. In the organism level, I innovated the computational methodology to assemble and analyze complete genomes of butterflies and discovered convergence evolution in butterfly wing patterns. In conclusion, I advanced the knowledge of biological sequences in multi-layers by computational approaches.