Alexander Kitaygorodsky

PhD student (DBMI), Summer 2016 - Spring 2021
University of Toronto

I am a PhD student in Dr. Yufeng Shen’s lab, at Columbia University’s Department of Biomedical Informatics. I am interested in the relationship between genetics and human disease, as well as interpretation of next-generation sequencing data with advanced machine learning techniques to gain genetic insights and to determine disease-causing variations. My thesis research is about integrating genetics, informatics, and machine learning to investigate disruption of genomic sites critical for transcriptional and post-transcriptional regulation. Specifically, I focus on the role of non-protein coding mutations in birth defects such as congenital heart disease, congenital diaphragmatic hernia, and autism, as this class of variants and the details of gene expression regulation represent significant knowledge gaps in the current field of genetics.

I am currently working on designing nested deep neural networks to predict position-specific selection constraint and genetic effect of noncoding mutations by combining information from two sources: disruption of RNA binding protein (RBP) binding as the cause of the genetic effect, and regional depletion of variation in human population as the consequence of the genetic effect. The method can be used to identify deleterious rare or de novo noncoding variants in disease studies, improve knowledge of genetic mechanisms, and aid discovery of specific new genetic targets that could be critical in future medical treatments and prevention.

Prior to working with Dr. Shen, I completed my Bachelor of Science at the University of Toronto, studying computer science and molecular genetics, and working with both Drs. Elisabeth Tillier and Philip M. Kim. In addition to my current focus on leveraging machine learning for noncoding genetics, I have also previously worked on projects in several other areas of bioinformatics including prediction of harmful drug-drug and drug-protein combinations, coevolution analysis to discover novel protein-protein interactions, and calling and analysis of structural noncoding variants in birth defects.

  1. Predicting localized affinity of RNA binding proteins to transcripts with convolutional neural networks bioRxiv. 2021
  2. Genomic analyses implicate noncoding de novo variants in congenital heart disease Nature Genetics. 2020
  3. Whole Genome De Novo Variant Identification with FreeBayes and Neural Network Approaches bioRxiv. 2020
  4. De novo variants in congenital diaphragmatic hernia identify MYRF as a new syndrome and reveal genetic overlaps with other developmental disorders. PLoS genetics. 2018