BABEL enables cross-modality translation between multiomic profiles at single-cell resolution
Published in Proceedings of the National Academy of Sciences, 2021
Recommended citation: Wu, K.E., Yost, K.E., Chang, H.Y. and Zou, J., 2021. BABEL enables cross-modality translation between multiomic profiles at single-cell resolution. Proceedings of the National Academy of Sciences, 118(15), p.e2023070118. https://www.pnas.org/doi/abs/10.1073/pnas.2023070118
Simultaneously profiling of multiomic modalities is a grand challenge for single-cell biology. While there have been impressive technical innovations demonstrating feasibily of such multi-omic profiling methods, these methods tend to be expensive and time-consuming. We present BABEL, an autoencoder-like deep learning method that can impute single-cell RNA from single-cell ATAC profiles, and vice versa. Once trained, BABEL can augment single-modality single-cell experiments with predicted values for unmeasured data modalities. We demonstrate that these imputations are highly accurate when evaluating on cell types similar to BABEL’s training set, regardless of variation from artifacts like batch effects.