We reasoned that an alternative approach would be to leverage a multi-omic dataset as a bridge that can help to translate between disparate modalities. To perform this translation, we were inspired by the field of dictionary learning, a form of representation learning that is commonly utilized in image analysis and also genomics33-37. The goal of dictionary learning is to represent input data, a noisy image for example, in terms of individual elements. These elements, such as image patches, are called atoms and together comprise a dictionary. Reconstructing an image as a weighted linear combination of these atoms is an effective tool for denoising, and represents a transformation of the image dataset into a dictionary-defined space.
