Title: Computational Inference of Cancer-Specific Vulnerabilities in Clinical Samples​

Speaker: Professor Choi Jung Kyoon (Associate Professor, Dept. of Bio & Brain Eng., KAIST)

Date & Time: March 7^th (Wed), 4:30PM~5:45PM

Venue: YBS Bldg. (E16-1), Rm#207

Abstract: Current methods to profile cancer dependency are applicable to only in vitro cell culture. Here, we developed an in silico RNAi method to simulate the transcriptomic perturbations caused by the suppression of each gene. The perturbation profiles were used to train deep neural networks. Cancer-specific vulnerabilities were predicted for a variety of genes involved in cell growth and maintenance in general from tumor and matched normal transcriptomes of clinical samples. Acquired dependencies of tumors were observed in cases in which one allele was inactivated by point mutations or in association with oncogenic mutations. Nucleocytoplasmic transport by Ran GTPase was identified as a common vulnerability in tumors with overactive growth signaling. Vulnerability to loss of Ku70/80 was predicted for tumors that appear to rely on classical nonhomologous end joining for DNA repair. In conclusion, this approach may facilitate the discovery of precision therapeutic targets from gene expression and mutation data.