Post Content Post navigation Building a pre-trained model for single-cell multi-omics analysis beyond highly variable feature selection