Comparative Genomics of Protein Structure-Function in Networks and Disease Clinical exome sequencing is increasingly frequent but still of limited impact on patient care and personalized medicine. Key difficulties keeping these data from being translated into therapeutic plans are that in each patient harmful mutations are few and seemlessly blended into the vast background of harmless ones. As a result, disease-causing genes are difficult to spot. Also, even when imputable genes are found their functions are typically unknown. We now tackle these problems by merging two lines of investigation that both model the evolution and propagation of biological information as a smooth process. In protein structures, smoothing has improved predictions of evolutionary important sites, functions and substrates. In networks, it has led to function predictions without need of structural knowledge. And, with respect to the genotype-phenotype relationship, it has led to an equation for the harmful action of coding mutations on fitness.
Aim 1 now proposes to refine this equation to score the harm of coding mutations in order to uncover disease- causing genes and pathways. The focus will be on cancer given the ready availability of exome data from The Cancer Genome Atlas. In parallel, Aim 2 will develop new and general network-based techniques to decipher the role of proteins of interest.
Aim 3 will experimentally validate select predictions of cancer-causing genes by testing in head and neck and lung cancer cell lines whether they behave as expected of tumor suppressors and oncogenes. This work should yield innovative, formal analyses of the genotype to phenotype relationship with direct applications to cancer genomics; it will unite molecular evolution and population genetics, whilst enabling the clinical interpretation of genome variations and the discovery of cancer genes that determine morbidity and mortality and eventually lead to novel therapeutics.
In order to guide patient therapy, this work will address two key problems: to understand the harm that comes from the many genome mutations we each carry, and to identify the genetic basis of diseases. While personal sequencing of genomes is on the brink of becoming routine, so far we have not been able to spot which of the many mutations it uncovers are in fact health threats. Innovatively, our solutions draw on fundamental concepts from mathematics, computer science and biology. Encouragingly, early results show these new methods have direct applications to understand cancer risk and treatment options, such as in head and neck cancers. We now aim to refine these ideas and systematically test them experimentally towards effective and personalized therapy.
Showing the most recent 10 out of 68 publications
