Tianshou Zhou’s research while affiliated with Sun Yat-sen University and other places

Manifold learning techniques have emerged as crucial tools for uncovering latent patterns in high-dimensional single-cell data. However, most existing dimensionality reduction methods primarily rely on 2D visualization, which can distort true data relationships and fail to extract reliable biological information. Here, we present DTNE (diffusive topology neighbor embedding), a dimensionality reduction framework that faithfully approximates manifold distance to enhance cellular relationships and dynamics. DTNE constructs a manifold distance matrix using a modified personalized PageRank algorithm, thereby preserving topological structure while enabling diverse single-cell analyses. This approach facilitates distribution-based cellular relationship analysis, pseudotime inference, and clustering within a unified framework. Extensive benchmarking against mainstream algorithms on diverse datasets demonstrates DTNE’s superior performance in maintaining geodesic distances and revealing significant biological patterns. Our results establish DTNE as a powerful tool for high-dimensional data analysis in uncovering meaningful biological insights.

The evolution processes of complex systems carry key information in the systems’ functional properties. Applying machine learning algorithms, we demonstrate that the historical formation process of various networked complex systems can be extracted, including protein-protein interaction, ecology, and social network systems. The recovered evolution process has demonstrations of immense scientific values, such as interpreting the evolution of protein-protein interaction network, facilitating structure prediction, and particularly revealing the key co-evolution features of network structures such as preferential attachment, community structure, local clustering, degree-degree correlation that could not be explained collectively by previous theories. Intriguingly, we discover that for large networks, if the performance of the machine learning model is slightly better than a random guess on the pairwise order of links, reliable restoration of the overall network formation process can be achieved. This suggests that evolution history restoration is generally highly feasible on empirical networks.

Revealing how transcriptional bursting kinetics are genomically encoded is challenging because genome structures are stochastic at the organization level and are suggestively linked to gene transcription. To address this challenge, we develop a generic theoretical framework that integrates chromatin dynamics, enhancer–promoter (E-P) communication, and gene-state switching to study transcriptional bursting. The theory predicts that power law can be a general rule to quantitatively describe bursting modulations by E-P spatial communication. Specifically, burst frequency and burst size are up-regulated by E-P communication strength, following power laws with positive exponents. Analysis of the scaling exponents further reveals that burst frequency is preferentially regulated. Bursting kinetics are down-regulated by E-P genomic distance with negative power-law exponents, and this negative modulation desensitizes at large distances. The mutual information between burst frequency (or burst size) and E-P spatial distance further reveals essential characteristics of the information transfer from E-P communication to transcriptional bursting kinetics. These findings, which are in agreement with experimental observations, not only reveal fundamental principles of E-P communication in transcriptional bursting but also are essential for understanding cellular decision-making.

Due to the conciseness and efficiency, percolation theory is widely applied in spreading dynamics which is a common and sophisticated phenomenon in real life. With the development of information technology, the quality and quantity of available data are improving, which offers the chance to describe and understand the empirical spreading phenomenon more comprehensively and accurately. Meanwhile, complicated dynamics brought by massive data pose new challenges to the research on social contagion based on percolation theory. In this prospective, we use several examples to briefly show how to use the percolation theory to describe information transmission on social networks driven by big data, to explore the indirect influence mechanism behind the spread of scientific research behavior, and to propose a new algorithm to quantify the global influence of nodes from the local topology. Finally, based on these studies, we propose several possible new directions of percolation theory in research of social contagion driven by big data.

Gene-expression bimodality, as a potential mechanism generating phenotypic cell diversity, can enhance the survival of cells in a fluctuating environment. Previous studies have shown that intrinsic or extrinsic regulations could induce bimodal gene expressions, but it is unclear whether this bimodality can occur without regulation. Here we develop an interpretable and tractable model, namely a generalized telegraph model (GTM), which considers silent transcription intervals and translational bursting, each being characterized by a general distribution. Using the queuing theory, we derive the analytical expressions of protein distributions, and show that non-exponential inactive times and translational bursting can lead to two peaks of the protein distribution away from the origin, which are different from those occurring in classical telegraph models. We also find that both silent-interval noise and translational burst-size noise can amplify gene-expression noise and induce diverse dynamic expression patterns. Our results not only provide an alternative mechanism of phenotypic switching but also could be used in explaining the bimodal phenomenon in experimental observations.

The study of resilience for high-dimensional systems is a challenging problem in physical sciences. Although a generic dimensional reduction method has been proposed to study the resilience of large-scale networks, an important yet unsolved question is how different types of regulation affect network resilience. Here we propose a new method to reduce the size of a large-scale regulatory network to the number of regulation types in the system. For each type of regulation, a symbolic variable is introduced to represent the function of that regulation in determining the resilience of the high-dimensional network. Using the genetic networks in Escherichia coli and Saccharomyces cerevisiae as the test systems, we examine how positive and negative regulation affect the resilience of these systems. Analytical and numerical studies suggest that the reduction of positive regulation deteriorates network resilience, while the decline of negative regulation enhances the resilience property. More importantly, we show that there is a trade-off between network efficiency and resilience, which is supported by the balance of positive and negative regulation. The proposed method provides a general framework to evaluate the key role of different regulatory mechanisms in determining the resilience of high-dimensional networks.

Background Single-cell technologies provide unprecedented opportunities to study heterogeneity of molecular mechanisms. In particular, single-cell RNA-sequence data have been successfully used to infer gene regulatory networks with stochastic expressions. However, there are still substantial challenges in measuring the relationships between genes and selecting the important genetic regulations. Objective This prospective provides a brief review of effective methods for the inference of gene regulatory networks. Methods We concentrate on two types of inference methods, namely the model-free methods and mechanistic methods for constructing gene networks. Results For the model-free methods, we mainly discuss two issues, namely the measures for quantifying gene relationship and criteria for selecting significant connections between genes. The issue for mechanistic methods is different mathematical models to describe genetic regulations accurately. Conclusions We and advocates the development of ensemble methods that combine two or more methods together.

Background: A major challenge in transcriptional regulation is to understand how static promoter architecture and dynamic feedback regulation dictate transcriptional burst kinetics on a genome-wide scale. Although single-cell RNA sequencing (scRNA-seq) provides an opportunity to address this issue, analytical methods need to be developed to uncover molecular mechanisms governing burst kinetics. Results: We develop an interpretable and scalable statistical framework, which combines experimental data with a mechanistic model to infer transcriptional burst kinetics (sizes and frequencies) and feedback regulations. Applying this framework to the scRNA-seq data of embryonic mouse fibroblast cells, we find that genome-wide burst kinetics exhibit different characteristics in cases without and with feedback regulations, implying Simpson’s paradoxes. We show that feedbacks modulate burst frequencies and sizes differently and conceal the effects of transcription start site distributions on burst kinetics. Notably, only in the presence of positive feedback, TATA genes are expressed with high burst frequencies, and enhancer-promoter interactions mainly modulate burst frequencies. Conclusions: The inference method developed here provides a flexible, robust, and efficient way to infer molecular mechanisms dictating burst kinetics. Our study reveals genome-wide burst patterns coordinated by promoter architecture and feedback regulations.

Although multistability is an important dynamic property of a wide range of complex systems, it is still a challenge to develop mathematical models for realising high order multistability using realistic regulatory mechanisms. To address this issue, we propose a robust method to develop multistable mathematical models by embedding bistable models together. Using the GATA1-GATA2-PU.1 module in hematopoiesis as the test system, we first develop a tristable model based on two bistable models without any high cooperative coefficients, and then modify the tristable model based on experimentally determined mechanisms. The modified model successfully realises four stable steady states and accurately reflects a recent experimental observation showing four transcriptional states. In addition, we develop a stochastic model, and stochastic simulations successfully realise the experimental observations in single cells. These results suggest that the proposed method is a general approach to develop mathematical models for realising multistability and heterogeneity in complex systems.

Gene expression in individual cells is inherently variable and sporadic, leading to cell-to-cell variability in mRNA and protein levels. Recent single-cell and single-molecule experiments indicate that promoter architecture and translational bursting play significant roles in controlling gene expression noise and generating the phenotypic diversity that life exhibits. To quantitatively understand the impact of these factors, it is essential to construct an accurate mathematical description of stochastic gene expression and find the exact analytical results, which is a formidable task. Here, we develop a stochastic model of bursty gene expression, which considers the complex promoter architecture governing the variability in mRNA expression and a general distribution characterizing translational burst. We derive the analytical expression for the corresponding protein steady-state distribution and all moment statistics of protein counts. We show that the total protein noise can be decomposed into three parts: the low-copy noise of protein due to probabilistic individual birth and death events, the noise due to stochastic switching between promoter states, and the noise resulting from translational busting. The theoretical results derived provide quantitative insights into the biochemical mechanisms of stochastic gene expression.