Mohammad Hosein Rohban

Assistant professor

Functional annotation of human genes and alleles using image-based profiling

  Halls department, Hall 5
  Thursday, 27 December 2018
  14:00 - 15:00


Image-based profiling has proven to be a powerful, efficient single-cell technology for characterizing the function of small molecules and genes at large scale. In image-based – or morphological – profiling, each cell population perturbed by a genetic or small molecule reagent is measured for a pattern – or signature – of the perturbants' effect on cell state. We recently used image-based profiling to systematically annotate functions of genes and alleles that are involved in various signaling pathways. Clustering these genes based on their profile similarities yields groups that, in most cases, correspond to biological pathways. The study revealed TRAF2/c-REL negative regulation of YAP1/WWTR1-responsive pathways, which was then confirmed at the transcriptional level.
The same approach can be used to assess the impact of various alleles of a gene in human disease, particularly for somatic mutations in cancer. We studied mutations that were observed in lung adenocarcinoma, and used overexpression constructs to study their impacts. We observe image-based profiling can predict the impact of gene mutations as accurately as gene expression profiling does, with the benefit of providing single-cell resolution, and adding complementary ways to interpret allele function and improve predictions.
Finally, in a preliminary study, we observed that small-molecule compounds whose image-based profile correlates (or anti-correlates) with the profile of a particular gene are more likely to be effective regulators of that gene. It would be revolutionary to be able to identify novel small molecule regulators for various genes based simply on matching their image-based profiles.


Mohammad Hossein Rohban is an assistant professor in Computer Engineering Department at Sharif University of Technology. His current research interests include Interpretable Machine Learning, Latent Variable Learning, and Computational Biology. In particular, he recently focused on Image-based profiling in high throughput assays using fluorescent microscopy, where images of cells are translated into fixed dimensional vectors, named as "profiles". His researches on Image-based profiling of small molecules and genetic perturbations at Broad Institute of Harvard and MIT led to identification of novel mechanistic relationships between signaling pathways involved in cancer and also discovery of novel small molecules that are potential regulators of these pathways.