scLinguist documentation

scLinguist a transformer-based framework for RNA-to-protein prediction that follows a two-step training strategy. First, we pretrain modality-specific models on large-scale single-omics datasets using self-supervised learning to extract informative representations. Subsequently, we fine-tune the model on paired RNA-protein data, enabling accurate cross-modality translation. This training paradigm allows our model to leverage abundant single-omics data while effectively learning modality relationships from limited paired datasets.

Overview of scLinguist workflow

Getting started with scLinguist

To begin using scLinguist, please refer to the following sections of the documentation:

• The Installation provides instructions for setting up scLinguist in your environment.

• The Tutorials contains examples on how to use scLinguist.