SCING: Inference of robust, interpretable gene regulatory networks from single cell and spatial transcriptomics

Russell Littman; Michael Cheng; Ning Wang; Chao Peng; Xia Yang

doi:10.1016/j.isci.2023.107124

SCING: Inference of robust, interpretable gene regulatory networks from single cell and spatial transcriptomics

iScience. 2023 Jun 14;26(7):107124. doi: 10.1016/j.isci.2023.107124. eCollection 2023 Jul 21.

Authors

Russell Littman^{1

2}, Michael Cheng^{1

2}, Ning Wang¹, Chao Peng³, Xia Yang^{1

2

4

5

6}

Affiliations

¹ Department of Integrative Biology & Physiology, UCLA, Los Angeles, CA, USA.
² Bioinformatics Interdepartmental Program, UCLA, Los Angeles, CA, USA.
³ Department of Neurology, David Geffen School of Medicine, UCLA, Los Angeles, CA, USA.
⁴ Institute for Quantitative and Computational Biosciences (QCBio), Los Angeles, CA, USA.
⁵ Molecular Biology Institute (MBI), Los Angeles, CA, USA.
⁶ Brain Research Institute (BRI), Los Angeles, CA, USA.

Abstract

Gene regulatory network (GRN) inference is an integral part of understanding physiology and disease. Single cell/nuclei RNA-seq (scRNA-seq/snRNA-seq) data has been used to elucidate cell-type GRNs; however, the accuracy and speed of current scRNAseq-based GRN approaches are suboptimal. Here, we present Single Cell INtegrative Gene regulatory network inference (SCING), a gradient boosting and mutual information-based approach for identifying robust GRNs from scRNA-seq, snRNA-seq, and spatial transcriptomics data. Performance evaluation using Perturb-seq datasets, held-out data, and the mouse cell atlas combined with the DisGeNET database demonstrates the improved accuracy and biological interpretability of SCING compared to existing methods. We applied SCING to the entire mouse single cell atlas, human Alzheimer's disease (AD), and mouse AD spatial transcriptomics. SCING GRNs reveal unique disease subnetwork modeling capabilities, have intrinsic capacity to correct for batch effects, retrieve disease relevant genes and pathways, and are informative on spatial specificity of disease pathogenesis.

Keywords: Biocomputational method; Machine learning; Transcriptomics.

Abstract

Grants and funding