Integrating unmatched scRNA-seq and scATAC-seq data and learning cross-modality relationship simultaneously

It is a challenging task to integrate scRNA-seq and scATAC-seq data obtained from different batches. Existing methods tend to use a pre-defined gene activity matrix (GAM) to convert the scATAC-seq data into scRNA-seq data. The pre-defined GAM is often of low quality and does not reflect the dataset-specific relationship between the two data modalities. We propose scDART ( s ingle c ell D eep learning model for A TAC-seq and R NA-seq T rajectory), a deep learning framework that integrates scRNA-seq and scATAC-seq data and learns cross-modalities relationships simultaneously. Specifically, the design of scDART allows it to preserve cell trajectories in continuous cell populations and can be applied to trajectory inference on integrated data.