Systematic mapping of organism-scale gene-regulatory networks in aging using population asynchrony.

In aging, physiologic networks decline in function at rates that differ between individuals, producing a wide distribution of lifespan. Though 70% of human lifespan variance remains unexplained by heritable factors, little is known about the intrinsic sources of physiologic heterogeneity in aging. To understand how complex physiologic networks generate lifespan variation, new methods are needed. Here, we present Asynch-seq, an approach that uses gene-expression heterogeneity within isogenic populations to study the processes generating lifespan variation. By collecting thousands of single-individual transcriptomes, we capture the Caenorhabditis elegans "pan-transcriptome"-a highly resolved atlas of non-genetic variation. We use our atlas to guide a large-scale perturbation screen that identifies the decoupling of total mRNA content between germline and soma as the largest source of physiologic heterogeneity in aging, driven by pleiotropic genes whose knockdown dramatically reduces lifespan variance. Our work demonstrates how systematic mapping of physiologic heterogeneity can be applied to reduce inter-individual disparities in aging.

Controlling organoid symmetry breaking uncovers an excitable system underlying human axial elongation.

The human embryo breaks symmetry to form the anterior-posterior axis of the body. As the embryo elongates along this axis, progenitors in the tail bud give rise to tissues that generate spinal cord, skeleton, and musculature. This raises the question of how the embryo achieves axial elongation and patterning. While ethics necessitate in vitro studies, the variability of organoid systems has hindered mechanistic insights. Here, we developed a bioengineering and machine learning framework that optimizes organoid symmetry breaking by tuning their spatial coupling. This framework enabled reproducible generation of axially elongating organoids, each possessing a tail bud and neural tube. We discovered that an excitable system composed of WNT/FGF signaling drives elongation by inducing a neuromesodermal progenitor-like signaling center. We discovered that instabilities in the excitable system are suppressed by secreted WNT inhibitors. Absence of these inhibitors led to ectopic tail buds and branches. Our results identify mechanisms governing stable human axial elongation.

A cross-disorder dosage sensitivity map of the human genome.

Rare copy-number variants (rCNVs) include deletions and duplications that occur infrequently in the global human population and can confer substantial risk for disease. In this study, we aimed to quantify the properties of haploinsufficiency (i.e., deletion intolerance) and triplosensitivity (i.e., duplication intolerance) throughout the human genome. We harmonized and meta-analyzed rCNVs from nearly one million individuals to construct a genome-wide catalog of dosage sensitivity across 54 disorders, which defined 163 dosage sensitive segments associated with at least one disorder. These segments were typically gene dense and often harbored dominant dosage sensitive driver genes, which we were able to prioritize using statistical fine-mapping. Finally, we designed an ensemble machine-learning model to predict probabilities of dosage sensitivity (pHaplo & pTriplo) for all autosomal genes, which identified 2,987 haploinsufficient and 1,559 triplosensitive genes, including 648 that were uniquely triplosensitive. This dosage sensitivity resource will provide broad utility for human disease research and clinical genetics.

Genomic structure predicts metabolite dynamics in microbial communities.

The metabolic activities of microbial communities play a defining role in the evolution and persistence of life on Earth, driving redox reactions that give rise to global biogeochemical cycles. Community metabolism emerges from a hierarchy of processes, including gene expression, ecological interactions, and environmental factors. In wild communities, gene content is correlated with environmental context, but predicting metabolite dynamics from genomes remains elusive. Here, we show, for the process of denitrification, that metabolite dynamics of a community are predictable from the genes each member of the community possesses. A simple linear regression reveals a sparse and generalizable mapping from gene content to metabolite dynamics for genomically diverse bacteria. A consumer-resource model correctly predicts community metabolite dynamics from single-strain phenotypes. Our results demonstrate that the conserved impacts of metabolic genes can predict community metabolite dynamics, enabling the prediction of metabolite dynamics from metagenomes, designing denitrifying communities, and discovering how genome evolution impacts metabolism.

Time-resolved systems immunology reveals a late juncture linked to fatal COVID-19.

COVID-19 exhibits extensive patient-to-patient heterogeneity. To link immune response variation to disease severity and outcome over time, we longitudinally assessed circulating proteins as well as 188 surface protein markers, transcriptome, and T cell receptor sequence simultaneously in single peripheral immune cells from COVID-19 patients. Conditional-independence network analysis revealed primary correlates of disease severity, including gene expression signatures of apoptosis in plasmacytoid dendritic cells and attenuated inflammation but increased fatty acid metabolism in CD56dimCD16hi NK cells linked positively to circulating interleukin (IL)-15. CD8+ T cell activation was apparent without signs of exhaustion. Although cellular inflammation was depressed in severe patients early after hospitalization, it became elevated by days 17-23 post symptom onset, suggestive of a late wave of inflammatory responses. Furthermore, circulating protein trajectories at this time were divergent between and predictive of recovery versus fatal outcomes. Our findings stress the importance of timing in the analysis, clinical monitoring, and therapeutic intervention of COVID-19.

Insertions and Deletions Target Lineage-Defining Genes in Human Cancers.

Certain cell types function as factories, secreting large quantities of one or more proteins that are central to the physiology of the respective organ. Examples include surfactant proteins in lung alveoli, albumin in liver parenchyma, and lipase in the stomach lining. Whole-genome sequencing analysis of lung adenocarcinomas revealed noncoding somatic mutational hotspots near VMP1/MIR21 and indel hotspots in surfactant protein genes (SFTPA1, SFTPB, and SFTPC). Extrapolation to other solid cancers demonstrated highly recurrent and tumor-type-specific indel hotspots targeting the noncoding regions of highly expressed genes defining certain secretory cellular lineages: albumin (ALB) in liver carcinoma, gastric lipase (LIPF) in stomach carcinoma, and thyroglobulin (TG) in thyroid carcinoma. The sequence contexts of indels targeting lineage-defining genes were significantly enriched in the AATAATD DNA motif and specific chromatin contexts, including H3K27ac and H3K36me3. Our findings illuminate a prevalent and hitherto unrecognized mutational process linking cellular lineage and cancer.

Structured States of Disordered Proteins from Genomic Sequences.

Protein flexibility ranges from simple hinge movements to functional disorder. Around half of all human proteins contain apparently disordered regions with little 3D or functional information, and many of these proteins are associated with disease. Building on the evolutionary couplings approach previously successful in predicting 3D states of ordered proteins and RNA, we developed a method to predict the potential for ordered states for all apparently disordered proteins with sufficiently rich evolutionary information. The approach is highly accurate (79%) for residue interactions as tested in more than 60 known disordered regions captured in a bound or specific condition. Assessing the potential for structure of more than 1,000 apparently disordered regions of human proteins reveals a continuum of structural order with at least 50% with clear propensity for three- or two-dimensional states. Co-evolutionary constraints reveal hitherto unseen structures of functional importance in apparently disordered proteins.

Investigating the potential of single-cell DNA methylation data to detect allele-specific methylation and imprinting.

Allele-specific methylation (ASM) is an epigenetic modification whereby one parental allele becomes methylated and the other unmethylated at a specific locus. ASM is most often driven by the presence of nearby heterozygous variants that influence methylation, but also occurs somatically in the context of genomic imprinting. In this study, we investigate ASM using publicly available single-cell reduced representation bisulfite sequencing (scRRBS) data on 608 B cells sampled from six healthy B cell samples and 1,230 cells from 11 chronic lymphocytic leukemia (CLL) samples. We developed a likelihood-based criterion to test whether a CpG exhibited ASM, based on the distributions of methylated and unmethylated reads both within and across cells. Applying our likelihood ratio test, 65,998 CpG sites exhibited ASM in healthy B cell samples according to a Bonferroni criterion (p < 8.4 × 10-9), and 32,862 CpG sites exhibited ASM in CLL samples (p < 8.5 × 10-9). We also called ASM at the sample level. To evaluate the accuracy of our method, we called heterozygous variants from the scRRBS data, which enabled variant-based calls of ASM within each cell. Comparing sample-level ASM calls to the variant-based measures of ASM, we observed a positive predictive value of 76%-100% across samples. We observed high concordance of ASM across samples and an overrepresentation of ASM in previously reported imprinted genes and genes with imprinting binding motifs. Our study demonstrates that single-cell bisulfite sequencing is a potentially powerful tool to investigate ASM, especially as studies expand to increase the number of samples and cells sequenced.

Semi-supervised machine learning method for predicting homogeneous ancestry groups to assess Hardy-Weinberg equilibrium in diverse whole-genome sequencing studies.

Large-scale, multi-ethnic whole-genome sequencing (WGS) studies, such as the National Human Genome Research Institute Genome Sequencing Program's Centers for Common Disease Genomics (CCDG), play an important role in increasing diversity for genetic research. Before performing association analyses, assessing Hardy-Weinberg equilibrium (HWE) is a crucial step in quality control procedures to remove low quality variants and ensure valid downstream analyses. Diverse WGS studies contain ancestrally heterogeneous samples; however, commonly used HWE methods assume that the samples are homogeneous. Therefore, directly applying these to the whole dataset can yield statistically invalid results. To account for this heterogeneity, HWE can be tested on subsets of samples that have genetically homogeneous ancestries and the results aggregated at each variant. To facilitate valid HWE subset testing, we developed a semi-supervised learning approach that predicts homogeneous ancestries based on the genotype. This method provides a convenient tool for estimating HWE in the presence of population structure and missing self-reported race and ethnicities in diverse WGS studies. In addition, assessing HWE within the homogeneous ancestries provides reliable HWE estimates that will directly benefit downstream analyses, including association analyses in WGS studies. We applied our proposed method on the CCDG dataset, predicting homogeneous genetic ancestry groups for 60,545 multi-ethnic WGS samples to assess HWE within each group.

KnockoffHybrid: A knockoff framework for hybrid analysis of trio and population designs in genome-wide association studies.

Both trio and population designs are popular study designs for identifying risk genetic variants in genome-wide association studies (GWASs). The trio design, as a family-based design, is robust to confounding due to population structure, whereas the population design is often more powerful due to larger sample sizes. Here, we propose KnockoffHybrid, a knockoff-based statistical method for hybrid analysis of both the trio and population designs. KnockoffHybrid provides a unified framework that brings together the advantages of both designs and produces powerful hybrid analysis while controlling the false discovery rate (FDR) in the presence of linkage disequilibrium and population structure. Furthermore, KnockoffHybrid has the flexibility to leverage different types of summary statistics for hybrid analyses, including expression quantitative trait loci (eQTL) and GWAS summary statistics. We demonstrate in simulations that KnockoffHybrid offers power gains over non-hybrid methods for the trio and population designs with the same number of cases while controlling the FDR with complex correlation among variants and population structure among subjects. In hybrid analyses of three trio cohorts for autism spectrum disorders (ASDs) from the Autism Speaks MSSNG, Autism Sequencing Consortium, and Autism Genome Project with GWAS summary statistics from the iPSYCH project and eQTL summary statistics from the MetaBrain project, KnockoffHybrid outperforms conventional methods by replicating several known risk genes for ASDs and identifying additional associations with variants in other genes, including the PRAME family genes involved in axon guidance and which may act as common targets for human speech/language evolution and related disorders.

EnhancerNet: a predictive model of cell identity dynamics through enhancer selection.

Understanding how cell identity is encoded by the genome and acquired during differentiation is a central challenge in cell biology. I have developed a theoretical framework called EnhancerNet, which models the regulation of cell identity through the lens of transcription factor-enhancer interactions. I demonstrate that autoregulation in these interactions imposes a constraint on the model, resulting in simplified dynamics that can be parameterized from observed cell identities. Despite its simplicity, EnhancerNet recapitulates a broad range of experimental observations on cell identity dynamics, including enhancer selection, cell fate induction, hierarchical differentiation through multipotent progenitor states and direct reprogramming by transcription factor overexpression. The model makes specific quantitative predictions, reproducing known reprogramming recipes and the complex haematopoietic differentiation hierarchy without fitting unobserved parameters. EnhancerNet provides insights into how new cell types could evolve and highlights the functional importance of distal regulatory elements with dynamic chromatin in multicellular evolution.

Cross-prediction-powered inference.

While reliable data-driven decision-making hinges on high-quality labeled data, the acquisition of quality labels often involves laborious human annotations or slow and expensive scientific measurements. Machine learning is becoming an appealing alternative as sophisticated predictive techniques are being used to quickly and cheaply produce large amounts of predicted labels; e.g., predicted protein structures are used to supplement experimentally derived structures, predictions of socioeconomic indicators from satellite imagery are used to supplement accurate survey data, and so on. Since predictions are imperfect and potentially biased, this practice brings into question the validity of downstream inferences. We introduce cross-prediction: a method for valid inference powered by machine learning. With a small labeled dataset and a large unlabeled dataset, cross-prediction imputes the missing labels via machine learning and applies a form of debiasing to remedy the prediction inaccuracies. The resulting inferences achieve the desired error probability and are more powerful than those that only leverage the labeled data. Closely related is the recent proposal of prediction-powered inference [A. N. Angelopoulos, S. Bates, C. Fannjiang, M. I. Jordan, T. Zrnic, Science 382, 669-674 (2023)], which assumes that a good pretrained model is already available. We show that cross-prediction is consistently more powerful than an adaptation of prediction-powered inference in which a fraction of the labeled data is split off and used to train the model. Finally, we observe that cross-prediction gives more stable conclusions than its competitors; its CIs typically have significantly lower variability.

The visual effect of wind turbines on property values is small and diminishing in space and time.

Renewable power generation is the key to decarbonizing the electricity system. Wind power is the fastest-growing renewable source of electricity in the United States. However, expanding wind capacity often faces local opposition, partly due to a perceived visual disamenity from large wind turbines. Here, we provide a US-wide assessment of the externality costs of wind power generation through the visibility impact on property values. To this end, we create a database on wind turbine visibility, combining information on the site and height of each utility-scale turbine having fed power into the U.S. grid, with a high-resolution elevation map to account for the underlying topography of the landscape. Building on hedonic valuation theory, we statistically estimate the impact of wind turbine visibility on home values, informed by data from the majority of home sales in the United States since 1997. We find that on average, wind turbine visibility negatively affects home values in an economically and statistically significant way in close proximity ([Formula: see text]5 miles/8 km). However, the effect diminishes over time and in distance and is indistinguishable from zero for larger distances and toward the end of our sample.

Explaining neural scaling laws.

The population loss of trained deep neural networks often follows precise power-law scaling relations with either the size of the training dataset or the number of parameters in the network. We propose a theory that explains the origins of and connects these scaling laws. We identify variance-limited and resolution-limited scaling behavior for both dataset and model size, for a total of four scaling regimes. The variance-limited scaling follows simply from the existence of a well-behaved infinite data or infinite width limit, while the resolution-limited regime can be explained by positing that models are effectively resolving a smooth data manifold. In the large width limit, this can be equivalently obtained from the spectrum of certain kernels, and we present evidence that large width and large dataset resolution-limited scaling exponents are related by a duality. We exhibit all four scaling regimes in the controlled setting of large random feature and pretrained models and test the predictions empirically on a range of standard architectures and datasets. We also observe several empirical relationships between datasets and scaling exponents under modifications of task and architecture aspect ratio. Our work provides a taxonomy for classifying different scaling regimes, underscores that there can be different mechanisms driving improvements in loss, and lends insight into the microscopic origin and relationships between scaling exponents.

A Markovian dynamics for Caenorhabditis elegans behavior across scales.

How do we capture the breadth of behavior in animal movement, from rapid body twitches to aging? Using high-resolution videos of the nematode worm Caenorhabditis elegans, we show that a single dynamics connects posture-scale fluctuations with trajectory diffusion and longer-lived behavioral states. We take short posture sequences as an instantaneous behavioral measure, fixing the sequence length for maximal prediction. Within the space of posture sequences, we construct a fine-scale, maximum entropy partition so that transitions among microstates define a high-fidelity Markov model, which we also use as a means of principled coarse-graining. We translate these dynamics into movement using resistive force theory, capturing the statistical properties of foraging trajectories. Predictive across scales, we leverage the longest-lived eigenvectors of the inferred Markov chain to perform a top-down subdivision of the worm's foraging behavior, revealing both "runs-and-pirouettes" as well as previously uncharacterized finer-scale behaviors. We use our model to investigate the relevance of these fine-scale behaviors for foraging success, recovering a trade-off between local and global search strategies.

Quantifying chemical short-range order in metallic alloys.

Metallic alloys often form phases-known as solid solutions-in which chemical elements are spread out on the same crystal lattice in an almost random manner. The tendency of certain chemical motifs to be more common than others is known as chemical short-range order (SRO), and it has received substantial consideration in alloys with multiple chemical elements present in large concentrations due to their extreme configurational complexity (e.g., high-entropy alloys). SRO renders solid solutions "slightly less random than completely random," which is a physically intuitive picture, but not easily quantifiable due to the sheer number of possible chemical motifs and their subtle spatial distribution on the lattice. Here, we present a multiscale method to predict and quantify the SRO state of an alloy with atomic resolution, incorporating machine learning techniques to bridge the gap between electronic-structure calculations and the characteristic length scale of SRO. The result is an approach capable of predicting SRO length scale in agreement with experimental measurements while comprehensively correlating SRO with fundamental quantities such as local lattice distortions. This work advances the quantitative understanding of solid-solution phases, paving the way for the rigorous incorporation of SRO length scales into predictive mechanical and thermodynamic models.

Adaptive nonequilibrium design of actin-based metamaterials: Fundamental and practical limits of control.

The adaptive and surprising emergent properties of biological materials self-assembled in far-from-equilibrium environments serve as an inspiration for efforts to design nanomaterials. In particular, controlling the conditions of self-assembly can modulate material properties, but there is no systematic understanding of either how to parameterize external control or how controllable a given material can be. Here, we demonstrate that branched actin networks can be encoded with metamaterial properties by dynamically controlling the applied force under which they grow and that the protocols can be selected using multi-task reinforcement learning. These actin networks have tunable responses over a large dynamic range depending on the chosen external protocol, providing a pathway to encoding "memory" within these structures. Interestingly, we obtain a bound that relates the dissipation rate and the rate of "encoding" that gives insight into the constraints on control-both physical and information theoretical. Taken together, these results emphasize the utility and necessity of nonequilibrium control for designing self-assembled nanostructures.

Nonreciprocity and odd viscosity in chiral active fluids.

Odd viscosity couples stress to strain rate in a dissipationless way. It has been studied in plasmas under magnetic fields, superfluid [Formula: see text], quantum-Hall fluids, and recently in the context of chiral active matter. In most of these studies, odd terms in the viscosity obey Onsager reciprocal relations. Although this is expected in equilibrium systems, it is not obvious that Onsager relations hold in active materials. By directly coarse-graining the kinetic energy and independently using both the Poisson-bracket formalism and a kinetic theory derivation, we find that the appearance of a nonvanishing angular momentum density, which is a hallmark of chiral active materials, necessarily breaks Onsager reciprocal relations. This leads to a non-Hermitian dynamical matrix for the total hydrodynamic momentum and to the appearance of odd viscosity and other nondissipative contributions to the viscosity. Furthermore, by accounting for both the angular momentum density and interactions that lead to odd viscosity, we find regions in the parameter space in which 3D odd mechanical waves propagate and regions in which they are mechanically unstable. The lines separating these regions are continuous lines of exceptional points, suggesting a possible nonreciprocal phase transition.

Homophily modulates double descent generalization in graph convolution networks.

Graph neural networks (GNNs) excel in modeling relational data such as biological, social, and transportation networks, but the underpinnings of their success are not well understood. Traditional complexity measures from statistical learning theory fail to account for observed phenomena like the double descent or the impact of relational semantics on generalization error. Motivated by experimental observations of "transductive" double descent in key networks and datasets, we use analytical tools from statistical physics and random matrix theory to precisely characterize generalization in simple graph convolution networks on the contextual stochastic block model. Our results illuminate the nuances of learning on homophilic versus heterophilic data and predict double descent whose existence in GNNs has been questioned by recent work. We show how risk is shaped by the interplay between the graph noise, feature noise, and the number of training labels. Our findings apply beyond stylized models, capturing qualitative trends in real-world GNNs and datasets. As a case in point, we use our analytic insights to improve performance of state-of-the-art graph convolution networks on heterophilic datasets.

Nanoscale dynamics of the cadherin-catenin complex bound to vinculin revealed by neutron spin echo spectroscopy.

We report a neutron spin echo (NSE) study of the nanoscale dynamics of the cell-cell adhesion cadherin-catenin complex bound to vinculin. Our measurements and theoretical physics analyses of the NSE data reveal that the dynamics of full-length α-catenin, ß-catenin, and vinculin residing in the cadherin-catenin-vinculin complex become activated, involving nanoscale motions in this complex. The cadherin-catenin complex is the central component of the cell-cell adherens junction (AJ) and is fundamental to embryogenesis, tissue wound healing, neuronal plasticity, cancer metastasis, and cardiovascular health and disease. A highly dynamic cadherin-catenin-vinculin complex provides the molecular dynamics basis for the flexibility and elasticity that are necessary for the AJs to function as force transducers. Our theoretical physics analysis provides a way to elucidate these driving nanoscale motions within the complex without requiring large-scale numerical simulations, providing insights not accessible by other techniques. We propose a three-way "motorman" entropic spring model for the dynamic cadherin-catenin-vinculin complex, which allows the complex to function as a flexible and elastic force transducer.