Towards Prediction of Phenotype from Genotype

Predicting phenotype from genotype represents the epitome of biological questions. As a multiscale problem, it starts from predicting exons and culminates with modeling of whole organisms. Focusing on the molecular level, I studied the relationship between sequences and protein spatial structures and analyzed proteins with similar sequences but different structures. To aid the assessment of structure prediction, I developed a method to rank the predictions of proteins with new folds, a very challenging problem that was previously addressed by expert inspection. Then, I developed a set of computer programs and scripts to predict various structural and functional properties of proteins from their sequences and implemented them as a public web-server. I applied these methods to important agricultural (citrus disease) and medical (Ebolavirus) problems. Moving on to organismal level predictions, I sequenced, annotated and analyzed complete genomes of butterflies and suggested hypotheses about genetic determinants of their behavior and other phenotypic traits. Taken together, these applications highlight the achievements possible today and challenges that lie ahead.