Dr. Qingrun Zhang

• Representation Learning

• Discover associations and causal factors in  multi-scale omics data

   

The high-throughput sequencing has dramatically reduced the cost of sequencing from $2.7 billion (in the Human Genome Project in 2003) to a few thousand dollars now; therefore it is widely believed that the fields of genetics, agriculture, and medicine will be revolutionized. On the path to achieve this exciting goal, there are many challenges. We have been working on the field focusing three fundamental questions: (1) How to discover genomic variants from the massive data and effectively store and analyze them using state-of-the-art computational techniques. (2) How to carry out statistical inference despite of the uncertainty brought by the sequencing and alignment errors. (3) How to identify associations between causal genes and the relevant phenotype in the presence of many confounding factors as well as gene-gene interactions using data mining models.

Currently, we are using genomics for cancer driver genes discovery and risk predictions. We also develop computational tools tailored to different cancer genomes.