Welcome to Pengyu Hong's Homepage

Image-based Functional Genomics

Image-based cellular phenotype data is a critical new modality of high-throughput data for functional genomics research. Studies of cellular phenotypes under various chemical stimulations and systematic perturbations in pathways by RNAi-based gene silencing will shed new light on the underlying biological networks, which lead various extra-cellular stimuli to numerous cellular phenotypes. Until now, cellular image analyses require highly trained scientists to spend days and months examining images and produce only a handful of qualitative results. Development of automatic image analysis and classification tools will significantly increase the feasibility and throughput of image-based gene function studies. I am developing image processing and pattern recognition techniques to extract image features and classify complex cellular morphology. This is a challenging research topic because of the great diversity of phenotypes and the high complexity of images due to the tendency of cells to clump together and overlap each other. At the same time, the highly dynamic range of morphology exhibited by RNAi-treated cells contains rich and profound biological information about gene functions yet to be discovered.

Some Preliminary Results

My method is capable of detecting both distinct features and subtle features that can barely be detected by naked eyes! Click to see images at higher resolution.

Quantitative image features can then be extracted for describing cell phenotypes.