Over my eight years at NYU in the Gunsalus-Piano lab, I have had the opportunity to work on a myriad of projects ranging from the detection of differential expression of transcript splice isoforms to novel network clustering tools to phenotype predictions in mouse early embryogenesis. All of my projects have shared the common feature of integrating multiple datatypes both to overcome the challenges of incomplete data and to identify emergent properties of systems analysis.
My most recent work was to develop a method that identifies genes that may be important for successful in vitro fertilization. We have approached this problem by looking to a mouse model system and developing an ensemble classification pipeline that leverages a variety of machine-learning methods. We have not only been able to predict phenotypes in worm, fly, human, yeast and mouse, but to combine each of these species-specific predictions–by using their mouse orthologs–and to generate a set of high-confidence genes that may give rise to early embryonic phenotyes in mice.