Understanding cellular responses to chemical interventions is critical to the discovery of effective therapeutics. Because individual biological techniques often measure only one axis of cellular response at a time, high-quality multimodal datasets are needed to unlock a holistic understanding of how cells respond to treatments and to advance computational methods that integrate modalities. However, many techniques destroy cells and thus preclude paired measurements, and attempts to match disparate unimodal datasets are often confounded by data being generated in incompatible experimental settings. Here we introduce scGeneScope, a multimodal single‑cell RNA sequencing (scRNA-seq) and Cell Painting microscopy image dataset conditionally paired by chemical treatment, designed to facilitate the development and benchmarking of unimodal, multimodal, and multiple profile machine learning methods for cellular profiling. 28 chemicals, each acting on distinct biological pathways or mechanisms of action (MoAs), were applied to U2-OS cells in two experimental data generation rounds, creating paired sets of replicates that were then profiled independently by scRNA‑seq or Cell Painting. Using scGeneScope, we derive a replicate- and experiment-split treatment identification benchmark simulating MoA discovery under realistic laboratory variability conditions and evaluate unimodal, multimodal, and multiprofile models ranging in complexity from linear approaches to recent foundation models. Multiprofile integration improved performance in both the unimodal and multimodal settings, with gains more consistent in the former. Evaluation of unimodal models for MoA identification demonstrated that recent scRNA-seq foundation models deployed zero-shot were consistently outperformed by classic fit-to-data methods, underscoring the need for careful, realistic benchmarking in machine learning for biology. We release the scGeneScope dataset and benchmarking code to support further research.

Communication is essential for biological and technological systems alike. Generally, increasing the volume of transmitted information is a priority for communication systems design. However, what if communication in living systems, which are subject to energy and noise constraints, is reliant on efficiently communicating "precise" information as opposed to "more" information? This question motivates the concept of semantic information. This study explores the trade-off between maximizing information transfer versus selecting semantically important information transfer in Biomolecular Communication Channels (BioMC). Specifically, we assume that a biological system needs to receive information to survive. This defines a viability function. We assume that maximizing information transfer optimizes the viability function by providing the maximum information to the receiver. We also assume that reducing information exchange reduces resource consumption. With this formulation, we propose a data-driven algorithm that seeks a transmission strategy that optimizes the viability function while reducing the information exchange. We validate the proposed methodology on in-silico data. Overall, this study provides a new framework for investigating semantic information transfer in BioMC.

Stochastic fluctuations in signaling and gene expression limit the ability of cells to sense the state of their environment, transfer this information along cellular pathways, and respond to it with high precision. Mutual information is now often used to quantify the fidelity with which information is transmitted along a cellular pathway. Mutual information calculations from experimental data have mostly generated low values, suggesting that cells might have relatively low signal transmission fidelity. In this work, we demonstrate that mutual information calculations might be artificially lowered by cell-to-cell variability in both initial conditions and slowly fluctuating global factors across the population. We carry out our analysis computationally using a simple signaling pathway and demonstrate that in the presence of slow global fluctuations, every cell might have its own high information transmission capacity but that population averaging underestimates this value. We also construct a simple synthetic transcriptional network and demonstrate using experimental measurements coupled to computational modeling that its operation is dominated by slow global variability, and hence that its mutual information is underestimated by a population averaged calculation.

Modeling and analysis of complex biological networks presents a number of mathematical challenges. For the models to be useful from a biological standpoint, they must be systematically compared with data. Robustness is a key to biological understanding and proper feedback to guide experiments,including both the deterministic stability and performance properties of models in the presence of parametric uncertainties and their stochastic behavior in the presence of noise. In this paper, we present mathematical and algorithmic tools to address such questions for models that may be nonlinear, hybrid,and stochastic. These tools are rooted in solid mathematical theories, primarily from robust control and dynamical systems, but with important recent developments. They also have the potential for great practical relevance, which we explore through a series of biologically motivated examples.

Biological systems have evolved complex regulatory mechanisms, even in situations where much simpler designs seem to be sufficient for generating nominal functionality. Using module-based analysis coupled with rigorous mathematical comparisons, we propose that in analogy to control engineering architectures, the complexity of cellular systems and the presence of hierarchical modular structures can be attributed to the necessity of achieving robustness. We employ the Escherichia coli heat shock response system, a strongly conserved cellular mechanism, as an example to explore the design principles of such modular architectures. In the heat shock response system, the sigma-factor sigma32 is a central regulator that integrates multiple feedforward and feedback modules. Each of these modules provides a different type of robustness with its inherent tradeoffs in terms of transient response and efficiency. We demonstrate how the overall architecture of the system balances such tradeoffs. An extensive mathematical exploration nevertheless points to the existence of an array of alternative strategies for the existing heat shock response that could exhibit similar behavior. We therefore deduce that the evolutionary constraints facing the system might have steered its architecture toward one of many robustly functional solutions.

In the technological and engineering sciences feedback control methods have been studied and applied extensively to man-made systems. Recently, there has been an increasing interest as to the applicability of feedback and control theory to the life sciences. In a previous paper (H. El-Samad et al., "Calcium homeostasis: a feedback control point of view," Proceedings of the 2000 American Control Conference, Chicago, Il, June 2000), the applicability of control theory to the analysis of calcium homeostasis was investigated. A mathematical model for this homeostatic mechanism was derived. Here, the authors provide an overview of this model, and then use it to study a pathological condition of the calcium regulatory mechanism in dairy cows. The clinical condition studied is referred to as parturient paresis (milk fever) and is the result of severe hypocalcemia at the time of parturition. The authors show that their mathematical feedback model for calcium homeostasis exhibits dynamic behavior that can correspond to milk fever. They then present a rigorous mathematical result that guarantees the presence of milk fever-like dynamics for a range of model parameters.