SPIRAL: integrating and aligning spatially resolved transcriptomics data across different experiments, conditions, and technologies

Tiantian Guo; Zhiyuan Yuan; Yan Pan; Jiakang Wang; Fengling Chen; Michael Q Zhang; Xiangyu Li

doi:10.1186/s13059-023-03078-6

SPIRAL: integrating and aligning spatially resolved transcriptomics data across different experiments, conditions, and technologies

Genome Biol. 2023 Oct 20;24(1):241. doi: 10.1186/s13059-023-03078-6.

Authors

Tiantian Guo^#^{1

2}, Zhiyuan Yuan^#³, Yan Pan⁴, Jiakang Wang¹, Fengling Chen⁵, Michael Q Zhang⁶, Xiangyu Li⁷

Affiliations

¹ School of Software Engineering, Beijing Jiaotong University, Beijing, 100044, China.
² MOE Key Laboratory of Bioinformatics, Bioinformatics Division and Center for Synthetic & Systems Biology, BNRist, Department of Automation, Tsinghua University, Beijing, 100084, China.
³ Institute of Science and Technology for Brain-Inspired Intelligence, Center for Medical Research and Innovation, Shanghai Pudong Hospital, Fudan University Pudong Medical Center, MOE Key Laboratory of Computational Neuroscience and Brain-Inspired Intelligence, Fudan University, Shanghai, 200433, China.
⁴ School of Biomedical Sciences, LKS Faculty of Medicine, The University of Hong Kong, Hong Kong SAR, China.
⁵ Center for Stem Cell Biology and Regenerative Medicine, MOE Key Laboratory of Bioinformatics, School of Life Sciences, Tsinghua-Peking Center for Life Sciences, Tsinghua University, Beijing, 100084, China.
⁶ Department of Biological Sciences, Center for Systems Biology, The University of Texas, Richardson, TX, 75080-3021, USA. michael.zhang@utdallas.edu.
⁷ School of Software Engineering, Beijing Jiaotong University, Beijing, 100044, China. lixiangyu@bjtu.edu.cn.

^# Contributed equally.

Abstract

Properly integrating spatially resolved transcriptomics (SRT) generated from different batches into a unified gene-spatial coordinate system could enable the construction of a comprehensive spatial transcriptome atlas. Here, we propose SPIRAL, consisting of two consecutive modules: SPIRAL-integration, with graph domain adaptation-based data integration, and SPIRAL-alignment, with cluster-aware optimal transport-based coordination alignment. We verify SPIRAL with both synthetic and real SRT datasets. By encoding spatial correlations to gene expressions, SPIRAL-integration surpasses state-of-the-art methods in both batch effect removal and joint spatial domain identification. By aligning spots cluster-wise, SPIRAL-alignment achieves more accurate coordinate alignments than existing methods.

Publication types

Research Support, Non-U.S. Gov't

MeSH terms

Gene Expression Profiling*
Transcriptome*