LTMG: a novel statistical modeling of transcriptional expression states in single-cell RNA-Seq data

Abstract

A key challenge in modeling single-cell RNA-seq data is to capture the diversity of gene expression states regulated by different transcriptional regulatory inputs across individual cells, which is further complicated by largely observed zero and low expressions. We developed a left truncated mixture Gaussian (LTMG) model, from the kinetic relationships of the transcriptional regulatory inputs, mRNA metabolism and abundance in single cells. LTMG infers the expression multi-modalities across single cells, meanwhile, the dropouts and low expressions are treated as left truncated. We demonstrated that LTMG has significantly better goodness of fitting on an extensive number of scRNA-seq data, comparing to three other state-of-the-art models. Our biological assumption of the low non-zero expressions, rationality of the multimodality setting, and the capability of LTMG in extracting expression states specific to cell types or functions, are validated on independent experimental data sets. A differential gene expression test and a co-regulation module identification method are further developed. We experimentally validated that our differential expression test has higher sensitivity and specificity, compared with other five popular methods. The co-regulation analysis is capable of retrieving gene co-regulation modules corresponding to perturbed transcriptional regulations. A user-friendly R package with all the analysis power is available at https://github.com/zy26/LTMGSCA.

scREAD: A Single-Cell RNA-Seq Database for Alzheimer's Disease

“…CTSRs are defined as a group of genes, which receive similar regulatory signals in a specific cell type, hence tending to have similar expression patterns and share conserved motifs in this cell type ( Wan et al., 2019 ; Yang et al., 2019 ). A successful elucidation of CTSRs will substantially improve the identification of transcriptionally co-regulated gene modules, realistically allowing reliable prediction of global transcription regulation networks encoded in a specific cell type ( Ma et al., 2020b ; Xie et al., 2020 ).…”

scLM: automatic detection of consensus gene clusters across multiple single-cell datasets

“…To evaluate the performance of scLM, we benchmarked it against other methods, including LTMG [26], SCN [25], Seurat-wgcna, MNN-wgcna, and SCENIC [24]. Seurat-wgcna and MNN-wgcna refer to the co-expression analysis using WGCNA [22], following the batch correction by Seurat [46] or MNN [47].…”

“…Classical methods designed for analysis of bulk transcriptome data such as WGCNA [22] and clust [23] are not designed to account for the unique characteristics of scRNA-seq data. Some network-based approaches for single cell data, including SCENIC [24], Cell Specific Network (CSN) [25], and LTMG [26], could detect gene co-expression modules as part of the network reconstruction. However, these methods didn’t account for the technical noise and extrinsic variance among multiple samples.…”

A graph neural network model to estimate cell-wise metabolic flux using single cell RNA-seq data

“…For (2), the modules that are known a priori to carry little or no flux will be excluded from further analysis. Specifically, for a given scRNA-seq data, scFEA will first determine for all the genes whether they have an active expression state using our in-house Left Truncated Mixture Gaussian model [43] (see details in Methods). The default setting of scFEA considers a module is blocked, if the module becomes disconnected after removing the reactions whose associated genes do not have significantly non-zero expressions throughout all the cell,.…”

“…We applied our inhouse method, LTMG, to determine the expression status of each genes in each single cell. LTMG considers the multi-modality of the expression profile of each gene throughout all the single cells, by assuming that the gene's expression follows a mixture of suppressed state and activated states, as represented by the following likelihood function [49].…”