Harry Huang
Foundation Model for In Silico Gene Perturbations
This project aims to develop a large-scale deep-learning foundation model trained on Perturb-seq datasets, which pair CRISPR gene edits with single-cell RNA-sequencing snapshots. By learning the transcriptional shifts that follow each perturbation, the model infers quantitative rules linking individual genes to the broader regulatory network across diverse human cell types. The resulting in-silico engine can rapidly predict how silencing or modulating any gene will reshape cellular programs, allowing researchers to prioritize the most informative laboratory experiments. This data-driven workflow accelerates target identification and streamlines early drug-discovery pipelines, reducing experimental overhead while sharpening biological insight.
Message To Sponsor
Thank you for investing in undergraduate research through the URAP Summer Fellowship. Your generosity allows me to devote the summer to building a computational model that could guide faster discovery of gene regulation network and gene-based therapies. Beyond advancing the project, the fellowship gives me invaluable training at the intersection of genomics and AI. I’m deeply grateful for the chance to turn curiosity into discoveries that may one day improve our lives.