DeepDEP

DeepDEP is a deep learning model designed to predict cancer dependencies by integrating genomic profiles, addressing the challenge of linking these dependencies to the molecular composition of cancer cells. The model uses unsupervised pretraining to capture tumor genomic representations, outperforming conventional methods and validating its accuracy with independent datasets. DeepDEP also extends dependency maps with functional characterizations and synthetic essentiality assays, leading to the creation of the first pan-cancer synthetic dependency map of 8,000 tumors with clinical relevance.