Research Interests

The research interests of the Khurana lab fall under the broad categories of genomics, computational biology and systems biology. We use computational models and machine learning approaches to analyze and interpret the terabytes of genomic, epigenomic and transcriptomic data from cancer patients. Our goal is to identify the genetic and epigenetic changes that enable initial tumor growth, and resistance to treatment in later stages. Towards this goal, we also develop computational methods that enable us to detect and monitor tumor genomic and epigenomic changes in minimally invasive manner using cell-free DNA whole-genome sequencing, as well as methods that leverage digital histopathology to infer tumor molecular states and disease progression. We also develop strategies for treatment of the new subtypes of tumors that we have identified, for example stem cell-like castration-resistant prostate cancer, using targeted and immunotherapeutic approaches. Our current focus is on prostate and breast cancers.

The major research themes can be summarized as follows.

1. Functional interpretation of the non-coding cancer genome

One of the lab's foundational interests has been understanding how mutations outside protein-coding genes contribute to cancer.

This includes:

  • identifying non-coding driver mutations
  • predicting functional regulatory variants
  • interpreting structural variants affecting regulatory elements
  • modeling how non-coding alterations influence gene regulation

Rather than focusing solely on mutations in genes, we have emphasized that regulatory DNA, including enhancers, promoters, and other cis-regulatory elements, contains important cancer-driving alterations.

2. Chromatin accessibility and gene regulatory networks

The overarching concept is that chromatin accessibility provides a dynamic readout of tumor state that is often more informative than DNA mutations alone.

Major directions include:

  • inference of gene regulatory networks
  • discovery of therapeutic vulnerabilities using chromatin accessibility
  • development of computational methods to infer tumor chromatin states from cell-free DNA whole-genome sequencing
3. Digital Pathology and Spatial Omics

We leverage advances in artificial intelligence to integrate H&E whole-slide images with spatial transcriptomics for characterization of tumor heterogeneity and the spatial organization of the tumor microenvironment.

This theme includes:

  • analysis of spatial transcriptomics data for tumor heterogeneity
  • analysis of the metastatic niche to identify therapeutic vulnerabilities
  • development of computational methods that connect tissue morphology with underlying molecular programs