scGET: Predicting Cell Fate Transition During Early Embryonic Development by Single-Cell Graph Entropy

Abstract

During early embryonic development, cell fate commitment represents a critical transition or “tipping point” of embryonic differentiation, at which there is a drastic and qualitative shift of the cell populations. In this study, we presented a computational approach, scGET, to explore the gene–gene associations based on single-cell RNA sequencing (scRNA-seq) data for critical transition prediction. Specifically, by transforming the gene expression data to the local network entropy, the single-cell graph entropy (SGE) value quantitatively characterizes the stability and criticality of gene regulatory networks among cell populations and thus can be employed to detect the critical signal of cell fate or lineage commitment at the single-cell level. Being applied to five scRNA-seq datasets of embryonic differentiation, scGET accurately predicts all the impending cell fate transitions. After identifying the “dark genes” that are non-differentially expressed genes but sensitive to the SGE value, the underlying signaling mechanisms were revealed, suggesting that the synergy of dark genes and their downstream targets may play a key role in various cell development processes. The application in all five datasets demonstrates the effectiveness of scGET in analyzing scRNA-seq data from a network perspective and its potential to track the dynamics of cell differentiation. The source code of scGET is accessible at https://github.com/zhongjiayuna/scGET_Project.

Distance covariance entropy reveals primed states and bifurcation dynamics in single-cell RNA-Seq data

“… 42 Importantly, DICE displayed a clear transition-like behavior, unlike the individual TF variances which increased but in a mostly asynchronous manner ( Figure 6 E). We also compared DICE to BioTIP, 19 a tool designed to detect cell-fate transitions using the concept of a dynamical network biomarker (DNB), 17 , 18 which also relies on covariation patterns, albeit not just of TFs but of specific subsets of all genes. In line with this, BioTIP’s criticality index also displayed a clear transition, although at an earlier timepoint compared to DICE, whilst also displaying larger fluctuations during the timecourse ( Figures S12 A and S12B).…”

“…It will be interesting if future work were to perform a more comprehensive comparison to include other tipping point algorithms such as scGET. 18 …”

“…In the face of these difficulties, single-cell RNA-Seq (scRNA-Seq) data 10 and other single-cell omic data types 11 offer the unprecedented opportunity to study the underlying bifurcation dynamics from a more data-driven perspective, as exemplified by the development of numerous lineage-trajectory inference algorithms 12 , 13 and statistical methods to detect TFs controlling the cell-fate commitment process. 5 , 14 , 15 , 16 In principle, with scRNA-Seq data it may be possible to identify the full repertoire of relevant regulatory factors, 14 to more accurately pinpoint the bifurcation or tipping points underlying cell-fate transitions, 15 , 16 , 17 , 18 , 19 , 20 , 21 or to better characterize system-wide properties such as priming in pluri-or-multipotent cell populations. 5 , 6 However, challenges remain.…”

Identifying the critical states and dynamic network biomarkers of cancers based on network entropy

“…The recently proposed dynamical network biomarker (DNB) theory shows a possible way of detecting the criticality of complex disease by regarding the progression of a disease as a high-dimensional nonlinear dynamic system and the critical transition as the state shift at the bifurcation point [ 1 , 7 , 8 ]. The DNB method and its modified versions have been applied in a variety of biomedical fields to successfully detect the pre-disease state of metabolic syndromes [ 9 , 10 ], identify immune checkpoint blockades [ 11 ] and assess cell fate commitment [ 12 ]. However, the DNB theory is not suitable for the analysis of datasets with small sample sizes since it requires multiple samples at each time point to evaluate its three statistical conditions, which restricts its application for biological and clinical data.…”

Advanced Single-Cell Omics Technologies and Informatics Tools for Genomics, Proteomics, and Bioinformatics Analysis

“…Meanwhile, a webserver, GranatumX was also presented in this special issue, which enables biologists to access the latest single-cell bioinformatics methods in a web-based graphical environment [22] . On the other hand, as a method-oriented work, the computational model proposed by Zhong et al [23] can explore the gene–gene associations based on scRNA-seq data for critical transition prediction. Li and Li proposed the scLink method, which used statistical network modeling to understand the co-expression relationships among genes and construct sparse gene co-expression networks from single-cell gene expression data [24] .…”