Computational approaches to biological data with applications in image analysis, human variant prioritization, and metagenomics

Advances in technology have produced efficient and powerful scientific instruments for measuring biological phenomena. In particular, modern microscopes and nextgeneration sequencing machines produce data at such a rate that manual analysis is no longer practical or feasible for meaningful scientific inquiries. Thus, there is a great need for computational strategies to organize and analyze huge amounts of data produced by biological experiments. My work presents computational strategies and software solutions for application in image analysis, human variant prioritization, and metagenomics. The information content of images can be leveraged to answer an extremely broad spectrum of questions ranging from inquiries about basic biological processes to highly specific, application-driven inquiries like the efficacy of a pharmaceutical drug. Modern microscopes can produce images at a rate at which rigorous manual analysis is impossible. I have created software pipelines that automate image analysis in two specific applications domains. In addition, I discuss general image analysis strategies that can be applied to a wide variety of problems. There are tens of millions of known human genetic variants. Prioritizing human variants based on how likely they are to cause disease is of huge importance because of the potential impact on human health. Current variant prioritization methods are limited by their scope, efficiency, and accuracy. I present a variant prioritization method, the VAAST variant prioritizer, which is superior in its scope, efficiency, and accuracy to existing variant prioritization methods. The rise of next-generation sequencing enables huge quantities of sequence to be generated in a short period of time. No field of study has been affected by rapid sequencing more than metagenomics. Metagenomics, the genomic analysis of a population 􀁌v of microorganisms, has important implications for pathogen detection because metagenomics enables the culture-free detection of microorganisms. I have created Taxonomer, a comprehensive metagenomics pipeline that enables the real-time analysis of read datasets derived from environmental samples.