Spatially informed clustering, integration, and deconvolution of spatial transcriptomics with GraphST

• DOI: 10.1101/2022.08.02.502407
• Publication Date: 2022-08-03T19:05:14Z
• Authors: Yahui Long; Kok Siong Ang; Mengwei Li; Kian Long Kelvin ...; Raman Sethi; Chengwei Zhong; Hang Xu; Zhiwei Ong; Karishma Sachaphi...; Ao Chen; Zeng Li; Huazhu Fu; Min Wu; Hsiu Kim Lina Lim; Longqi Liu; Jinmiao Chen
• Keywords: 0301 basic medicine; 0303 health sciences; 03 medical and health sciences; Science; Gene Expression Profiling; Q; Brain; Cluster Analysis; Transcriptome; Germinal Center; Article

Abstract Advances in spatial transcriptomics technologies have enabled the gene expression profiling of tissues while retaining its context. Effective exploitation this data combination requires spatially informed analysis tools to perform three key tasks, clustering, multi-sample integration, and cell type deconvolution. Here, we present GraphST, a novel graph self-supervised contrastive learning method that incorporates location information profiles accomplish all tasks streamlined process outperforming existing methods each task. GraphST combines neural networks with learn informative discriminative spot representations by minimizing embedding distance between adjacent spots vice versa. With achieved 10% higher clustering accuracy on multiple datasets than competing methods, better delineated fine-grained structures such as brain embryo. Moreover, is only can jointly analyze tissue slices both vertical horizontal integration correcting for batch effects. Lastly, compared other GraphST’s deconvolution simulated captured niches germinal centers lymph node experimentally acquired data. We further showed recover immune distribution different regions breast tumor reveal exhausted infiltrating T cells. Through our examples, demonstrated widely applicable broad range types technology platforms. In summary, streamlined, user friendly computationally efficient tool characterizing complexity gaining biological insights into organization within tissues.