Single‐cell gene regulatory network analysis for mixed cell populations

Gene Regulatory Networks (GRNs) reveal how genes regulate each other, offering insights into complex biological systems. Traditional GRN inference, based on bulk RNA sequencing, reflects average gene expression across mixed cell populations, obscuring single-cell regulatory dynamics. Single-cell RNA sequencing (scRNA-seq) has transformed gene expression studies by enabling GRN inference at single-cell resolution. However, most methods assume a shared GRN across all cells, overlooking the distinct GRNs of different cell types. Traditionally, GRN inference for scRNA-seq with mixed cell types involves a two-step process: first, clustering cells to identify cell types, and second, inferring GRNs for each cell type. While effective for clearly separated cell types, this approach struggles when cell types are highly mixed, leading to misclassification and reduced accuracy in GRN inference.

Recently, Quantitative Biology published an approach entitled "Single-cell gene regulatory network analysis for mixed cell populations", introducing VMPLN, a novel method that uses a mixture Poisson log-normal model for scRNA-seq data. VMPLN combines clustering and network inference through variational inference, improving the accuracy and robustness of cell type-specific GRN analysis.

To infer gene regulatory networks (GRNs) of mixed populations in scRNA-seq data, the research team utilized the mixture Poisson lognormal (MPLN) model to analyze scRNA-seq count expression data (Figure 1). The precision matrices of the MPLN are the GRNs of different cell types. To avoid the intractable optimization of the MPLN’s log‐likelihood, an algorithm called variational mixture Poisson log‐normal (VMPLN) is proposed to jointly estimate the GRNs of different cell types based on the variational inference method. Comprehensive simulation shows that VMPLN outperforms state-of-the-art single-cell GRN inference methods, especially in scenarios where different cell types have a high mixing degree. Benchmarking on real scRNA‐seq data also demonstrates that VMPLN can provide more accurate network estimation in most cases. Through applying to a large scRNA‐seq dataset from patients infected with severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2), VMPLN identifies critical differences in regulatory networks in immune cells between patients with moderate and severe symptoms.