B0678
Title: Cell location recovery with CeLEry: A supervised deep-learning algorithm for discovering spatial origins in scRNA-seq
Authors: Qihuang Zhang - McGill University (Canada) [presenting]
Abstract: Single-cell RNA sequencing has revolutionized the understanding of cellular heterogeneity in health and disease. However, the lack of spatial relationships among dissociated cells has limited its applications. CeLEry, a supervised deep-learning algorithm to recover the spatial origins of cells in scRNA-seq by leveraging gene expression and spatial location relationships learned from spatial transcriptomics. CeLEry integrates an optional data augmentation procedure using a variational autoencoder, enhancing the method's robustness and addressing noise in scRNA-seq. Additionally, it employs a co-embedding process to extract common latent features from multiple modalities. CeLEry effectively infers spatial origins at various levels, including 2D location and the spatial domain of a cell. Comprehensive benchmarking evaluations on multiple datasets from brain and cancer tissues demonstrate CeLErys reliability in recovering spatial location information for cells in scRNA-seq.
