A pioneering study published in Nature Communications introduces MOSA (Multi-Omic Synthetic Augmentation), an unsupervised deep learning model designed to integrate and augment multi-omic datasets of cancer cell lines, significantly enhancing the understanding of cancer biology and drug response mechanisms.
Led by researchers from the Children’s Medical Research Institute and the Instituto Superior Técnico, the study presents MOSA as a powerful tool for generating molecular and phenotypic profiles, increasing the number of multi-omic profiles by 32.7% and providing a complete DepMap for 1,523 cancer cell lines. This synthetic augmentation of data not only boosts statistical power but also uncovers less studied mechanisms associated with drug resistance and refines the identification of genetic associations and clustering of cancer cell lines.
The MOSA model harnesses orthogonal multi-omic information, successfully generating comprehensive structurally informed protein interactomes. By applying SHapley Additive exPlanations (SHAP) for model interpretation, MOSA reveals multi-omic features essential for cell clustering and biomarker identification related to drug and gene dependencies. This understanding is crucial for developing effective strategies to prioritize cancer targets.
"MOSA represents a significant leap forward in our ability to predict and understand the impact of genetic variations on protein interactions," said Emanuel Gonçalves, co-lead author of the study. "This platform has the potential to significantly accelerate the translation of genetic findings into clinical applications."
The ability of MOSA to synthetically generate datasets that are missing in specific samples addresses the pervasive dataset gaps in well-characterized models such as cancer cell lines. This positions MOSA as a valuable tool for in silico testing and prioritization of drug targets for experimental validation. The MOSA model is implemented as both a web server platform and a software package, offering researchers a powerful tool to explore the functional genomics of diseases and perform on-demand interface predictions.
The study is a collaborative effort involving researchers from the Children’s Medical Research Institute, the Instituto Superior Técnico, and the Wellcome Sanger Institute. The work was supported by the Wellcome Trust and the European Union’s Horizon 2020 research and innovation program.
Original link https://www.nature.com/articles/s41467-024-54771-4
