Cell Systems

Predictive logical models have made substantial contributions to biomedical research by facilitating the identification of regulatory mechanisms driving biological processes. Yet, their construction remains a major bottleneck, as it typically relies on extensive prior knowledge at the genome scale. The emergence of high-throughput omics technologies has opened new opportunities for data-driven logical model construction by providing high-resolution observations of biological states. In parallel, formal constraint-based approaches have emerged, reformulating executable model inference as a constraint-satisfaction problem. However, transforming omics observations into biologically meaningful Boolean state abstractions and dynamical constraints suitable for logical synthesis engines remains challenging, particularly for poorly understood biological systems and multi-condition experimental settings. To address these challenges, we introduce scBOLT, a semi-automated framework for Boolean constraint engineering and logical model inference from multi-condition single-cell transcriptomes. Beyond these methodological contributions, scBOLT substantially improves computational tractability, enabling the inference of Boolean networks comprising several hundred components within practical runtimes.

Genetic perturbations can reshape cell populations by altering the relative abundance of specific cell types and states within the profiled population, including increases, decreases and states that become detectable after perturbation. Pooled single-cell screens, such as Perturb-seq, measure such responses at scale. However, only a small fraction of possible perturbations can be tested experimentally. A central challenge is therefore to predict compositional shifts induced by unseen perturbations. Many perturbation-prediction methods do not directly optimize for this outcome; instead, they predict gene-expression responses and infer cell-type and cell-state composition downstream. Surprisingly, we find that even models that accurately predict perturbation-induced changes in average gene expression perform poorly at forecasting these compositional shifts. To address this gap, we present Pop-Corn, a method that directly predicts how a perturbation reshapes cell-type composition without reconstructing gene expression. In the primary T-cell benchmark, Pop-Corn predicted the overall cell-state composition of held-out perturbations more accurately than the evaluated expression-prediction pipelines, while better preserving the diversity of observed cell states. We further extend Pop-Corn to intact tissue, where it predicts perturbation-induced cell-type proportion changes in local cellular neighborhoods and uses attention patterns to generate hypotheses about context-dependent cellular interactions. Retrospective virtual screens support the use of Pop-Corn to prioritize perturbations for experimental follow-up according to their predicted effects on cell-state composition.

Mechanistic models of biological networks predict but cannot be calibrated at scale, while learned representations describe but expose no dynamical operator. Both lack an inverse step mapping data onto the few dynamical quantities the data determine. We develop Unified Network Spectroscopy (UNS) as that operator and establish when it is well posed. Reading fluctuations about a physiological setpoint through the Lyapunov relation of the linear noise approximation, UNS returns dominant timescales and frequencies, the subspace on which they act, and the projected noise. The inversion itself is established; our contribution is architectural. Three results govern its success. First, biochemical noise is generically anisotropic, and extrinsic modulation makes the dynamics non-normal. Assuming isotropy can invert mode rankings and destroy precision-matrix topology recovery; we give the exact, checkable condition under which recovery survives. Second, equal-time covariance identifies neither the dynamics nor their spectrum: the residual freedom is precisely the probability current that non-equilibrium living systems generically carry. Third, population snapshots replace time series only under ergodicity; heterogeneity biases inferred relaxation rates downward, making systems appear closer to bifurcation, whatever the sample size. Lagged covariance or per-cell velocity restores identifiability and turns noise structure into a recovered, falsifiable quantity. We provide a three-regime ergodicity taxonomy with diagnostics and show that the variational operator remains well posed when samples are fewer than variables. The output is a typed interface contract for hierarchical model composition, carrying its uncertainty, attainment and validity domain, with controllability inheritance as a falsification test.

Microbiome measurements are commonly judged by how faithfully they reconstruct an underlying biological profile, using taxonomic similarity, prediction accuracy, or compositional distance. Yet faithfulness and usefulness are not the same: a distorted measurement may still support the correct intervention, while a small error can be decisive when it falls near the boundary between two treatment choices. We therefore evaluate measurement quality by whether information loss changes the decisions and outcomes that remain attainable. We treat the intervention-relevant state of the gut community as a hidden quantity seen only through an imperfect, information losing measurement, mapped by a decision rule to a graded intervention. Each decision rule is evaluated by net host benefit and ecological margin, generating a feasible set of attainable outcome pairs. We define the measurement-induced decision loss (MIDL) as the maximum net host benefit lost under imperfect measurement relative to perfect observation. For scalar affine information loss, increasing degradation is a Blackwell garbling; under post-processing closure, more informative measurement weakly expands the feasible set and cannot reduce optimal host benefit. In the binary-state model, MIDL increases linearly while an observation-dependent decision rule remains optimal and plateaus once the optimal policy becomes signal-independent. We extend the framework to finite states, state-specific loss, multi-segment stool aggregation, and ecological-margin constraints. The framework distinguishes measurement faithfulness from decision usefulness and provides a principled basis for comparing stool and richer measurements by the interventions and trade-offs they support.

Pooled Perturb-seq screens are often interpreted through a linear steady-state model in which each targeted gene supplies an additive perturbation input. Although this formulation identifies an operator on the observed response subspace in closed form, identifiability alone does not establish that inferred operators generalize to new perturbation targets. We evaluated fixed program-space additive inputs in three Perturb-seq screens -- Replogle K562 essential, Replogle RPE1 essential, and Jost 2020 -- using target-grouped nested cross-validation and matched signal-to-noise linear-truth controls. For inverse operator prediction, real held-out error remained at the predict-zero baseline (0.96, 1.00, and 1.00), whereas matched linear controls achieved 0.18, 0.53, and 0.72. This discrepancy persisted across random K562 target panels, an overdetermined Jost dimension sweep, direction-only fits, ridge and truncated-SVD estimators, and five linear ground-truth ensembles. Forward target-held-out prediction was more nuanced: fixed, footprint, and learned linear encodings modestly beat a predict-training-mean baseline in K562 but not in RPE1 or Jost, and the K562 gains recovered only about 7% of the gap to the matched linear control. On the tested screens, fitted fixed program-space operators should not be interpreted as validated target-generalizing regulatory operators. We release anchor-op, a reproducible toolkit for target-grouped evaluation, identifiability checks, and matched-geometry positive controls for perturbation-response models.

Virtual cell models aim to predict how cells respond to perturbations, yet accurate prediction of post-perturbation gene expression can conceal failures to recover context-specific responses. Here we identify context geometry compression (CGC), in which models preserve cellular identity and shared perturbation effects while collapsing differences in how the same perturbation acts across contexts. Across biological systems, reproducible context-specific responses remained difficult to recover; in a large drug-response matrix, this persisted even with nearly complete perturbation coverage. The same drug-response predictions achieved higher normalized recovery in predefined pathway summaries than in gene-level response profiles, while coarse response programs were more predictable in human lymphoblastoid cell lines. Compact baseline transcriptomic representations retained most full-RNA predictive performance for patient-specific ex vivo drug responses. Together, these findings establish recoverable biological resolution, rather than measurement dimensionality, as a criterion for evaluating virtual cell predictions and matching model outputs to the biological detail supported by evidence.

Identifying which gene drives a desired cell-state transition is a foundational inverse problem in target discovery and cell-state engineering, yet far less developed than forward prediction. We show that prevailing benchmarks inflate performance through two artefacts: training leakage, where test-perturbation cells are seen during fitting, and evaluation leakage, where the perturbed gene's own expression change exposes the answer. Here, we introduce Cause-excluded Effect Evaluation (CEE), a framework that separates train and test perturbation targets, removes target-gene self-effects from expression profiles and scores candidates in downstream response space. With CEE, we find that simple structure-based methods consistently outperform complex deep learning models across 16 single-cell perturbation datasets, revealing that the recoverable signal is governed by low-rank and additive structure in perturbation responses rather than by model complexity. Our work disentangles evaluation artefacts from genuine predictive capacity and offers a design principle for developing perturbation target identification methods.

Predicting cellular response to perturbations is key to understanding biological mechanisms and selecting therapeutic targets. However, generalizing across cellular contexts, perturbations, and experimental datasets remains challenging due to incomplete representations of the system being perturbed and the difficulty of measuring true perturbation effects within that system. We introduce PIE, a model that addresses these limitations by reformulating the learning task around population level perturbation effects and incorporating auxiliary inputs that better characterize the biological system. PIE directly predicts differentially expressed genes and changes in gene expression for each (context, perturbation) pair, using external biological knowledge, baseline gene expression, and observed perturbation responses. Its architecture accommodates input sources with variable feature sets and predicts effects on genes seen and unseen during training. On the Replogle Nadig dataset, PIE achieves state of the art performance on most metrics across all tested generalization settings, including the most challenging setting where both the context and perturbation are unseen during training. For predicting differentially expressed genes, PIE achieves 1.2 to 3.2 times the AUPRC of the strongest baseline in each setting: across unseen contexts, perturbations, experimental datasets, and jointly unseen contexts and perturbations. Overall, PIE provides a framework for generalizing perturbation effects across diverse experimental conditions.

Single-cell transcriptomics has transformed our ability to characterize cellular identity, but present-day gene expression profiles capture only a snapshot of a process that unfolds across cell division history. Recent single-cell lineage-tracing technologies make it possible to reconstruct cell division histories for thousands of cells, opening a window into how gene expression evolves. Yet observed gene expression is often redundant, with correlations among genes reflecting underlying latent biological programs and regulatory networks. To capture this structure, we introduce scPFA, a single-cell phylogenetic factor analysis framework that represents gene expression through a small set of latent factors that evolve along lineages under a phylogenetic prior, capturing correlated structure hidden in present-day observations alone. Simulations demonstrate accurate recovery of latent factors and covariance structure across conditions. Applied to developmental and cancer lineage-tracing datasets, the model uncovers biologically interpretable, lineage-associated expression programs. Together, these results demonstrate how lineage-informed factor analysis can reveal temporal biological structure hidden within high-dimensional single-cell data.

Cells employ dense networks of transient protein-protein interactions mediated by modular peptide-binding domains and unstructured peptidic motifs for high-fidelity information processing. How these networks physically execute computations through protein interactions governed by complex intra- and intermolecular mechanisms remains indiscernible from current, sparse and non-quantitative, maps of the human interactome. Here, we introduce a quantitative statistical mechanical modeling (QSM) approach for machine learning domain-peptide affinities with experimental-level accuracy. Leveraging a new, principled algorithm for data harmonization and a biophysically informed neural network architecture, QSM learns to predict dissociation constants directly from amino acid sequences with calibrated confidence. We use QSM to construct the first quantitative drafts of human signaling networks and study these networks across three physical scales--recognition mechanisms of modular binding domains, combinatorial logic of multi-dentate proteins, and pathways inferred from de novo inference of protein interaction networks. We find that (i) modular domains, based on their binding preferences, selectivities, and strengths, fall into a limited number of biophysical equivalence groups, (ii) those domains, along with peptidic motifs, are "syntactically" combined within proteins to yield multivalent recognition mechanisms, and (iii) the organization of cellular function can be traced back to algorithmically detectable modules induced by domain-mediated interactions. In aggregate, these analyses instantiate a tractable roadmap towards a comprehensive, mechanistic, and simulatable articulation of the systems biology of signaling.

Reliable biological prediction by AI is most likely to emerge first in the simplest systems with rich datasets and the fastest opportunities for testing and refinement. Phages with small genomes are an ideal testbed. Decades of experimental work have created an unusually information-rich literature benchmark that sits largely outside the sequence repositories typically used to train genome language models (GLMs). Recent work has shown that GLMs such as Evo2 can generate viable whole bacteriophage genomes, demonstrating that genome-scale biological design is possible. The next question is more mechanistic: can such models correctly predict the effects of simple, local sequence changes? Here, we evaluated Evo2 against experimental mutation data from two model phages: the single-stranded RNA phage MS2 and the single-stranded DNA phage {Phi}X174. We first addressed whether Evo2 sequence log-likelihood scores could distinguish viable from nonviable mutations, and then whether models trained on Evo2 embeddings were predictive of function. Across both phages, Evo2 captured expected sequence-level constraints: stop codons were generally penalized, synonymous substitutions had higher likelihood than nonsynonymous substitutions. Evo2 distinguished among synonymous codons in ways only weakly explained by host codon usage. However, for MS2, these capabilities did not translate into robust biological predictions. Specifically, Evo2 {Delta}SLL failed to distinguish functional from nonfunctional mutations in the lysis gene, an overlapping viral region that appears to be a particularly hard test case. In {Phi}X174, performance was stronger but still modest, and much of the apparent signal could be explained by simple covariates such as nonsense mutations, genomic position, and nucleotide distance from the reference. Together, these results introduce phages as a tractable proving ground for stress-testing GLMs against experimentally grounded genotype-to-phenotype tasks. More broadly, they provide a durable framework for identifying what data and model ingredients are required for biologically reliable prediction.

Relating the drivers of binding affinity to structural features remains difficult because affinity emerges from many weak, context-dependent interactions, while experimental affinity and mutation data are limited. Here, we ask whether a machine-learning model can learn energetically relevant interaction preferences directly from native structures, without affinity or mutation labels. We decompose interfaces into local structural motifs and learn the compatibility of each motif with its surrounding molecular environment. Applied across a protein surface, these models yield target preference maps (TPMs), spatial fields of local interaction compatibility. Trained exclusively on native antibody-antigen structures without affinity or mutation labels, TPMs recover charged, aromatic and backbone-mediated recognition across peptide-protein and other non-antibody interfaces. On an out-of-domain SKEMPI benchmark, wild-type TPM scores discriminated mutation-sensitive positions (AUC 0.643, n=1323), while TPM-based substitution scores ranked alternative amino acids comparably to FoldX and Rosetta Flex {Delta}{Delta}G. These results show that native interface structures can supervise learning of transferable, energetically relevant molecular preferences without direct thermodynamic labels.

From flocking birds to swarming insects, collective motion enables groups to accomplish tasks beyond the capabilities of individuals. These behaviors emerge from local interactions among group members, but the interaction rules that organize collectives at the cellular scale remain poorly understood, in part because they are difficult to manipulate experimentally. Here we develop a cell-machine interface that enables us to rewrite the rules of collective migration. Cell-cell interaction rules are specified in software and implemented in living cells through real-time computer vision and closed-loop optogenetic control. Using a single engineered cell line, we program cells to self-organize into flocks, to propagate regenerative signal relays that recapitulate the interaction logic of immune-cell swarms, and to collectively navigate guidance cues that individual cells cannot. Our work shows that altering simple rules of cell-cell interaction is sufficient to generate new emergent collective capabilities. This approach provides an experimental framework for discovering the organizational principles of multicellular coordination and designing synthetic collectives with capabilities beyond those of their individual members.

Scientific discovery seeks mechanisms that explain observations and predict beyond the measurements that produced them. Whereas large language models (LLMs) encode knowledge implicitly, mechanistic world models organise it as a parsimonious set of explicit, modular, reusable mechanisms whose predictions can be scored against data. In biology, where measurements are sparse and noisy, mechanistic world modelling must discover latent states and governing equations jointly, yet the prior knowledge that could constrain this search is largely unstructured. We introduce gemot, an agentic framework for mechanistic world modelling, and evaluate it on 18 published biological problems spanning molecular biology, epidemiology, and immune-cell differentiation, with up to 1,755 training measurements, 65 observables, 170 experimental conditions, and 3 data modalities per problem. gemot is auditable by construction: a semantic layer records each hypothesis, and a Model Context Protocol layer separates hypothesis formulation from numerical evaluation, so hypothesis scoring cannot be fabricated. On every problem, autonomously constructed models match or exceed the reference models in fit and parsimony, and generalise where held-outs exist. We demonstrate that gemot can formulate novel, biologically plausible mechanistic hypotheses. This work will enable decoding of interventional biological data into competing mechanistic hypotheses and design experiments that distinguish between them, and may serve as a blueprint for mechanistic world modelling in other domains.

Single-cell perturbation response prediction seeks to recover transcriptional responses under unseen perturbation conditions. Because RNA sequencing measurements are destructive, control and perturbed cells are typically observed as unpaired populations without cell-level correspondence, shifting the prediction objective from individual cellular outcomes to perturbation-specific population distributions. The Schr"odinger Bridge (SB) provides a principled framework for modeling these population-level transitions as stochastic transport, but learning bridge dynamics from high-dimensional, sparse gene-expression profiles remains challenging. We propose PerturbBridge, a conditional latent Schr\"odinger Bridge framework that reformulates stochastic population transport over high-dimensional, sparse gene-expression profiles as bridge learning in a compact cell latent space. This formulation enables an efficient approximation of SB-based stochastic population transport in single-cell perturbation prediction. PerturbBridge first learns a compact latent representation following the \beta-TCVAE paradigm \citep{chen2018isolating}, alleviating the challenges caused by high dimensionality and sparsity during bridge learning. An interpolation-consistency regularizer further encourages agreement between decoded latent interpolations and the corresponding expression-space interpolations. PerturbBridge then learns a perturbation-conditioned latent bridge using a tractable stochastic SB approximation between latent control and target populations, with endpoint MMD regularization promoting alignment between generated and observed target distributions. Experiments on the Norman CRISPR and Sci-Plex3 chemical perturbation benchmarks demonstrate competitive performance across all evaluation metrics. Notably, PerturbBridge achieves state-of-the-art performance in differential-expression recovery on both benchmarks, highlighting the effectiveness of latent stochastic transport modeling for population-level single-cell perturbation prediction.

High-content imaging-based screens with small molecule or genetic perturbations rely on rich optical phenotypes, yet how distinct subcellular markers contribute complementary biological information remains unclear. Here, we introduce CellMAPP, a multiplexed organelle imaging platform that uses optical pooled screens with iterative antibody staining to map cellular architecture under perturbation. Using 17 markers across 524 genetic perturbations in MCF-7 cells, we systematically examined how individual, and combinations of, markers capture diverse morphological responses. We find that markers provide complementary views of perturbation effects, with increased multiplexing improving signal and the breadth of hit recovery, although some phenotypes are better recovered from a subset of features rather than a full profile. Because non-molecular image-based profiles can be challenging to link with interpretable biology, we developed CHART (available at https://github.com/Genentech/CHART), an archetype analysis framework that identifies distinct reference morphological states for each marker and uses multimodal large language models (MLLMs) to automatically annotate them, connecting cellular images to recognizable organelle phenotypes. Integrating archetypes across multiple markers revealed modular morphological programs that describe coordinated organelle reorganization in response to perturbation. This framework identified regulators of defined organelle states such as micronuclei, nominated novel gene functions and revealed complex patterns of multi-organelle reposition. Together, optical profiling with MAPP and CHART provides an interpretable, scalable framework for image-based screens that sheds light on rules governing cellular structural reprogramming.

Spatial omics technologies provide unprecedented opportunities to investigate how molecular programs are organized within tissues. However, existing analytical frameworks primarily focus on identifying cellular states and characterizing differentially expressed molecular features across these states. The relational spatial organization of molecular features and its dynamics across biological contexts therefore remain largely unexplored. Here, we present GeneSPOT, a gene-centric computational framework for characterizing the spatial relationships among genes and quantifying their reorganization across biological conditions. Although gene-centric in design, GeneSPOT generalizes to other molecular features. Across diverse applications, GeneSPOT resolved tissue structures, integrated paired and unpaired spatial multi-omics data, unveiled the spatial convergence of gene programs during embryonic development, and identified aging-associated spatial reorganization among genes whose expression did not change significantly with age. GeneSPOT thus complements conventional cell-centric analysis, providing a framework for deciphering how spatial relationships among molecular features are reorganized across biological conditions and over time.

Predicting the transcriptomic consequences of cellular perturbations is challenging due to the sparsity of single-cell transcriptional responses, heterogeneous perturbation efficiency, and the difficulty of identifying the small subset of genes that are truly differentially expressed after perturbation. Current approaches typically encode interventions as latent shifts from unperturbed to perturbed cellular state, making the models hard to interpret. Here we introduce SPECTRA (SPEctral CRISPR Transcriptome Regulatory Autoencoder) a graph-based model that treats CRISPR perturbations as interpretable, localized signals injected into a Gene Regulatory Network and propagated through directed graph neural networks. SPECTRA combines a variational control-cell encoder and a directed heterophilic graph decoder, integrating single-cell expression, pretrained transcriptomic context, and prior regulatory topology into a single node-level representation. We evaluate SPECTRA on a large-scale single-cell CRISPRi benchmark, measuring differentially expressed gene recovery through precision, F1 score, and AUPRC-based DEG classification. Since perturbations in SPECTRA are interpretable, we tested the predictions of the model against known biology, in cases where the effects of a gene knock out have been repeatedly demonstrated with orthogonal experiments. Our results show that graph-based signal propagation is a biologically grounded alternative to latent-shift perturbation modeling and can improve the recovery of sparse perturbation-induced transcriptional effects.

Proteins are often classified as toxic or non-toxic without measuring the abundance reached, leaving constraints on tolerable protein abundance unresolved. We established a limit-pushing approach in Saccharomyces cerevisiae combining strong inducible expression with gTOW-mediated high-copy selection to counteract copy-number compensation while measuring protein abundance and growth. Nearly all of approximately 80 chromosome I proteins severely inhibited growth or reduced viability at sufficiently high abundance. We established IE50, the expression level associated with a 50% reduction in growth rate, to quantify their widely varying overexpression tolerance. IE50 was positively associated with predicted structural order and cytoplasmic localization propensity and negatively associated with sulphur content. Single-cell imaging linked higher tolerance to proteins remaining cytoplasmic without becoming aggregation-positive and revealed abundance-dependent changes in localization and organelle morphology. At extreme abundance, Fun12, Nup60, and Pex22 generated distinct large-scale intracellular states through specific sequence regions. These findings establish overexpression toxicity as a quantitative property linked to protein characteristics and reveal both constraints on tolerable abundance and sequence-dependent capacities for intracellular organization.

Volumetric cell segmentation is limited by the scarcity of labelled data, because manual curation of three-dimensional datasets is prohibitively laborious. We find that this gap can be bridged with volumes in which synthetic 3D objects mirror the properties of real cells and of image formation, notably the axial spread of the point-spread function. Such volumes provide unlimited labelled training data, and we show that a model trained on them produces accurate segmentations of real biological data.