eLife: latest articles by subject

Human-specific lncRNAs contributed critically to human evolution by distinctly regulating gene expression

zhuhao@smu.edu.cn (Hao Zhu) — Fri, 13 Mar 2026 00:00:00 +0000

What genes and regulatory sequences critically differentiate modern humans from apes and archaic humans, which share highly similar genomes but show distinct phenotypes, has puzzled researchers for decades. Previous studies examined species-specific protein-coding genes and related regulatory sequences, revealing that birth, loss, and changes in these genes and sequences drive speciation and evolution. However, investigations of species-specific lncRNA genes and related regulatory sequences, which regulate substantial genes, remain limited. We identified human-specific (HS) lncRNAs from GENCODE-annotated human lncRNAs, predicted their DNA-binding domains (DBDs) and DNA-binding sites (DBSs), analyzed DBS sequences in modern humans (CEU, CHB, and YRI), archaic humans (Altai Neanderthals, Denisovans, and Vindija Neanderthals), and chimpanzees, and investigated how HS lncRNAs and their DBSs have influenced gene expression in archaic and modern humans. Our results suggest that these lncRNAs and DBSs have substantially reshaped gene expression, and this reshaping has evolved continuously from archaic to modern humans, enabling humans to adapt to new environments and lifestyles, promoting brain evolution, and resulting in cross-population differences. The parallel analysis of gene expression in GTEx tissues by HS transcription factors (TFs) and their DBSs indicates that HS lncRNAs have reshaped gene expression in the brain more significantly than HS TFs.

Cell type-specific network analysis in Diversity Outbred mice identifies genes potentially responsible for human bone mineral density GWAS associations

crf2s@virginia.edu (Charles Farber) — Wed, 11 Mar 2026 00:00:00 +0000

Genome-wide association studies (GWASs) have identified many sources of genetic variation associated with bone mineral density (BMD), a clinical predictor of fracture risk and osteoporosis. Aside from the identification of causal genes, other difficult challenges to informing GWAS include characterizing the roles of predicted causal genes in disease and providing additional functional context, such as the cell-type predictions or biological pathways in which causal genes operate. Leveraging single-cell transcriptomics (scRNA-seq) can assist in informing BMD GWAS by linking disease-associated variants to genes and providing a cell-type context for which these causal genes drive disease. Here, we use large-scale scRNA-seq data from bone marrow-derived stromal cells cultured under osteogenic conditions (BMSC-OBs) from Diversity Outbred (DO) mice to generate cell type-specific networks and contextualize BMD GWAS-implicated genes. Using trajectories inferred from the scRNA-seq data that map cell state transitions, we identify networks enriched with genes that exhibit the most dynamic changes in expression across trajectories. We discover 21 network driver genes, which are likely to be causal for human BMD GWAS associations that colocalize with expression/splicing quantitative trait loci (eQTLs/sQTLs). These driver genes, including Fgfrl1 and Tpx2, along with their associated networks, are predicted to be novel regulators of BMD via their roles in the differentiation of mesenchymal lineage cells. In this work, we showcase the use of single-cell transcriptomics from mouse bone-relevant cells to inform human BMD GWAS and prioritize genetic targets with potential causal roles in the development of osteoporosis.

Automated genome mining predicts structural diversity and taxonomic distribution of peptide metallophores across bacteria

nadine.ziemert@uni-tuebingen.de (Alison Butler) — Mon, 09 Mar 2026 00:00:00 +0000

Microbial competition for trace metals shapes their communities and interactions with humans and plants. Many bacteria scavenge trace metals with metallophores, small molecules that chelate environmental metal ions. Metallophore production may be predicted by genome mining, where genomes are scanned for homologs of known biosynthetic gene clusters (BGCs). However, accurately detecting non-ribosomal peptide (NRP) metallophore biosynthesis requires expert manual inspection, stymieing large-scale investigations. Here, we introduce automated identification of NRP metallophore BGCs through a comprehensive algorithm, implemented in antiSMASH, that detects chelator biosynthesis genes with 97% precision and 78% recall against manual curation. We showcase the utility of the detection algorithm by experimentally characterizing metallophores from several taxa. High-throughput NRP metallophore BGC detection enabled metallophore detection across 69,929 genomes spanning the bacterial kingdom. We predict that 25% of all bacterial non-ribosomal peptide synthetases encode metallophore production and that significant chemical diversity remains undiscovered. A reconstructed evolutionary history of NRP metallophores supports that some chelating groups may predate the Great Oxygenation Event. The inclusion of NRP metallophore detection in antiSMASH will aid non-expert researchers and continue to facilitate large-scale investigations into metallophore biology.

Deep learning linking mechanistic models to single-cell transcriptomics data reveals transcriptional bursting in response to DNA damage

jiangbenyuan@gdph.org.cn (Benyuan Jiang) — Wed, 04 Mar 2026 00:00:00 +0000

Cells must adopt flexible regulatory strategies to make decisions regarding their fate, including differentiation, apoptosis, or survival in the face of various external stimuli. One key cellular strategy that enables these functions is stochastic gene expression programs. However, understanding how transcriptional bursting, and consequently, cell fate, responds to DNA damage on a genome-wide scale poses a challenge. In this study, we propose an interpretable and scalable inference framework, DeepTX, that leverages deep learning methods to connect mechanistic models and single-cell RNA sequencing (scRNA-seq) data, thereby revealing genome-wide transcriptional burst kinetics. This framework enables rapid and accurate solutions to transcription models and the inference of transcriptional burst kinetics from scRNA-seq data. Applying this framework to several scRNA-seq datasets of DNA-damaging drug treatments, we observed that fluctuations in transcriptional bursting induced by different drugs were associated with distinct fate decisions: 5′-iodo-2′-deoxyuridine treatment was associated with differentiation in mouse embryonic stem cells by increasing the burst size of gene expression, while low- and high-dose 5-fluorouracil treatments in human colon cancer cells were associated with changes in burst frequency that corresponded to apoptosis- and survival-related fate, respectively. Together, these results show that DeepTX enables genome-wide inference of transcriptional bursting from single-cell transcriptomics data and can generate hypotheses about how bursting dynamics relate to cell fate decisions.

JAX Animal Behavior System (JABS), a genetics-informed, end-to-end advanced behavioral phenotyping platform for the laboratory mouse

Vivek.Kumar@jax.org (Anshul Choudhary) — Mon, 02 Mar 2026 00:00:00 +0000

Automated detection of complex animal behavior remains a challenge in neuroscience. Developments in computer vision have greatly advanced automated behavior detection and allow high-throughput preclinical and mechanistic studies. An integrated hardware and software solution is necessary to facilitate the adoption of these advances in the field of behavioral neurogenetics, particularly for non-computational laboratories. We have published a series of papers using an open field arena to annotate complex behaviors such as grooming, posture, and gait as well as higher-level constructs such as biological age and pain. Here, we present our integrated rodent phenotyping platform, JAX Animal Behavior System (JABS), to the community for data acquisition, machine learning-based behavior annotation and classification, classifier sharing, and genetic analysis. The JABS Data Acquisition Module (JABS-DA) enables uniform data collection with its combination of 3D hardware designs and software for real-time monitoring and video data collection. JABS-Active Learning Module (JABS-AL) allows behavior annotation, classifier training, and validation. We introduce a novel graph-based framework (ethograph) that enables efficient boutwise comparison of JABS-AL classifiers. JABS-Analysis and Integration Module (JABS-AI), a web application, facilitates users to deploy and share any classifier that has been trained on JABS, reducing the effort required for behavior annotation. It supports the inference and sharing of the trained JABS classifiers and downstream genetic analyses (heritability and genetic correlation) on three curated datasets spanning 168 mouse strains that we are publicly releasing alongside this study. This enables the use of genetics as a guide to proper behavior classifier selection. This open-source tool is an ecosystem that allows the neuroscience and genetics community to share advanced behavior analysis and reduces the barrier to entry into this new field.

Raw signal segmentation for estimating RNA modification from Nanopore direct RNA sequencing data

lu.cheng.ac@gmail.com (Aki Vehtari) — Mon, 02 Mar 2026 00:00:00 +0000

Estimating RNA modifications from Nanopore direct RNA sequencing data is a critical task for the RNA research community. However, current computational methods often fail to deliver satisfactory results due to inaccurate segmentation of the raw signal. We have developed a new method, SegPore, which leverages a molecular jiggling translocation hypothesis to improve raw signal segmentation. SegPore is a pure white-box model with enhanced interpretability, significantly reducing structured noise in the raw signal. We demonstrate that SegPore outperforms state-of-the-art methods, such as Nanopolish and Tombo, in raw signal segmentation across three large benchmark datasets. Moreover, the improved signal segmentation achieved by SegPore enables SegPore+m6Anet to deliver state-of-the-art performance in site-level m6A identification. Additionally, SegPore surpasses baseline methods like CHEUI in single-molecule level m6A identification.

Agent-based modeling reveals how bats navigate dense group emergences

omer_mazar@yahoo.com (Omer Mazar) — Mon, 02 Mar 2026 00:00:00 +0000

Bats face a complex navigation challenge when emerging from densely populated roosts, where vast numbers take off at once in dark, confined spaces. Each bat must avoid collisions with walls and conspecifics while locating the exit, all amidst overlapping acoustic signals. This crowded environment creates the risk of acoustic jamming, in which the calls of neighboring bats interfere with echo detection, potentially obscuring vital information. Despite these challenges, bats navigate these conditions with remarkable success. Although bats have access to multiple sensory cues, here, we focused on whether echolocation alone could provide sufficient information for orientation under such high-interference conditions. To explore whether and how they manage this challenge, we developed a sensorimotor model that mimics the bats’ echolocation behavior under high-density conditions. Our model suggests that the problem of acoustic jamming may be less severe than previously assumed. Frequent calls with short inter-pulse intervals (IPI) increase the sensory input flow, allowing integration of echoic information across multiple calls. When combined with simple movement-guidance strategies—such as following walls and avoiding nearby obstacles—this accumulated information enables effective navigation in dense acoustic environments. Together, these findings demonstrate a plausible mechanism by which bats may overcome acoustic interference and underscore the role of signal redundancy in supporting robust echolocation-based navigation. Beyond advancing our understanding of bat behavior, they also offer valuable insights for swarm robotics and collective movement in complex environments.

Motor biases reflect a misalignment between visual and proprioceptive reference frames

shion07070017@gmail.com (Amber Jiang) — Thu, 19 Feb 2026 00:00:00 +0000

Goal-directed movements can fail due to errors in our perceptual and motor systems. While these errors may arise from random noise within these sources, they also reflect systematic motor biases that vary with the location of the target. The origin of these systematic biases remains controversial. Drawing on data from an extensive array of reaching tasks conducted over the past 30 years, we evaluated the merits of various computational models regarding the origin of motor biases. Contrary to previous theories, we show that motor biases produced by human participants do not arise from systematic errors associated with the sensed hand position during motor planning or from the biomechanical constraints imposed during motor execution. Rather, motor biases are primarily caused by a misalignment between eye-centric and body-centric representations of position. This model can account for motor biases across a wide range of contexts, encompassing movements with the right versus left hand, finger versus hand movements, visible and occluded starting positions, as well as before and after sensorimotor adaptation.

Ribosome demand links transcriptional bursts to protein expression noise

riddhiman.dhar@iitkgp.ac.in (Riddhiman Dhar) — Wed, 18 Feb 2026 00:00:00 +0000

Stochastic variation in protein expression generates phenotypic heterogeneity in a cell population and has an important role in antibiotic persistence, mutation penetrance, tumor growth, and therapy resistance. Studies investigating molecular origins of noise have predominantly focused on the transcription process. However, the noise generated in the transcription process is further modulated by translation. This influences the expression noise at the protein level which eventually determines the extent of phenotypic heterogeneity in a cell population. Studies across different organisms have revealed a positive association between translational efficiency and protein noise. However, the molecular basis of this association has remained unknown. In this work, through stochastic modeling of translation in single mRNA molecules and empirical measurements of protein noise, we show that ribosome demand associated with high translational efficiency in a gene drives the correlation between translational efficiency and protein noise. We also show that this correlation is present only in genes with bursty transcription. Thus, our work reveals the molecular basis of how coding sequence of genes, along with their promoters, can regulate noise. These findings have important implications for investigating protein noise and phenotypic heterogeneity across biological systems.

Linking spinal circuit reorganization to recovery after thoracic spinal cord injury

smd395@drexel.edu (Andrew B Lockhart) — Tue, 17 Feb 2026 00:00:00 +0000

Rats exhibit significant recovery of locomotor function following incomplete spinal cord injuries, albeit with altered gait expression and reduced speed and stepping frequency. These changes likely result from and give insight into the reorganization within spared and injured spinal circuitry. Previously, we developed computational models of the mouse spinal locomotor circuitry controlling speed-dependent gait expression (Danner et al., 2017; Zhang et al., 2022). Here, we adapted these models to the rat and used the adapted model to explore potential circuit-level changes underlying altered gait expression observed after recovery from two different thoracic spinal cord injuries (lateral hemisection and contusion) that have roughly comparable levels of locomotor recovery (Danner et al., 2023). The model reproduced experimentally observed gait expression before injury and after recovery from lateral hemisection and contusion and suggests two distinct, injury-specific routes to restored function. First, recovery after lateral hemisection required substantial functional restoration of damaged descending drive and long propriospinal connections, suggesting compensatory plasticity through formation of detour pathways. Second, recovery after a moderate midline contusion predominantly relied on reorganization of spared sublesional networks and altered control of supralesional cervical circuits, compensating for weakened propriospinal and descending pathways. Despite these differences, sensitivity analysis revealed that restored activation of sublesional lumbar rhythm-generating circuits and appropriately balanced lumbar commissural connectivity are the key determinants of post-injury gait expression, suggesting that injury symmetry shapes how the cord reorganizes, but effective recovery in both cases depends on re-engaging these lumbar networks, which makes them prime targets for therapeutic intervention.

Deep neural networks to register and annotate cells in moving and deforming nervous systems

flavell@mit.edu (Adam A Atanas) — Tue, 17 Feb 2026 00:00:00 +0000

Aligning and annotating the heterogeneous cell types that make up complex cellular tissues remains a major challenge in the analysis of biomedical imaging data. Here, we present a series of deep neural networks that allow for automatic non-rigid registration and cell identification, developed in the context of freely moving and deforming invertebrate nervous systems. A semi-supervised learning approach was used to train a Caenorhabditis elegans registration network (BrainAlignNet) that aligns pairs of images of the bending C. elegans head with single-pixel-level accuracy. When incorporated into an image analysis pipeline, this network can link neurons over time with 99.6% accuracy. This network could also be readily purposed to align neurons from the jellyfish Clytia hemisphaerica, an organism with a vastly different body plan and set of movements. A separate network (AutoCellLabeler) was trained to annotate >100 neuronal cell types in the C. elegans head based on multi-spectral fluorescence of genetic markers. This network labels >100 different cell types per animal with 98% accuracy, exceeding individual human labeler performance by aggregating knowledge across manually labeled datasets. Finally, we trained a third network (CellDiscoveryNet) to perform unsupervised discovery of >100 cell types in the C. elegans nervous system: by comparing multi-spectral imaging data from many animals, it can automatically identify and annotate cell types without using any human labels. The performance of CellDiscoveryNet matched that of trained human labelers. These tools should be immediately useful for a wide range of biological applications and should be straightforward to generalize to many other contexts requiring alignment and annotation of dense heterogeneous cell types in complex tissues.

Single-cell atlas of AML reveals age-related gene regulatory networks in t(8;21) AML

georges.lacaud@manchester.ac.uk (Georges Lacaud) — Wed, 11 Feb 2026 00:00:00 +0000

Acute myeloid leukemia (AML) is characterized by cellular and genetic heterogeneity, which correlates with clinical course. Although single-cell RNA sequencing (scRNA-seq) reflects this diversity to some extent, the low sample numbers in individual studies limit the analytic potential when comparing specific patient groups. We performed large-scale integration of published scRNA-seq datasets to create a unique single-cell transcriptomic atlas for AML (AML scAtlas), totaling 748,679 cells, from 159 AML patients and 51 healthy donors from 20 different studies. This is the largest single-cell data resource for human AML to our knowledge, publicly available at https://cellxgene.cziscience.com/collections/071b706a-7ea7-47a4-bddf-6457725839fc. This AML scAtlas allowed investigations into 20 patients with t(8;21) AML, where we explored the clinical importance of age, given the in-utero origin of pediatric disease. We uncovered age-associated gene regulatory network (GRN) signatures, which we validated using bulk RNA sequencing data to delineate distinct groups with divergent biological characteristics. Furthermore, using an additional multiomic dataset (scRNA-seq and scATAC-seq), we validated our initial findings and created a de-noised enhancer-driven GRN reflecting the previously defined age-related signatures. Applying integrated data analysis of the AML scAtlas, we reveal age-dependent gene regulation in t(8;21) AML, potentially reflecting immature/fetal HSC origin in prenatal origin disease vs postnatal origin. Our analysis revealed that BCLAF1, which is particularly enriched in pediatric AML with t(8;21) of inferred in-utero origin, is a promising prognostic indicator. The AML scAtlas provides a powerful resource to investigate molecular mechanisms underlying different AML subtypes.

The representation of facial emotion expands from sensory to prefrontal cortex with development

bijanki@bcm.edu (Abhishek Tripathi) — Fri, 30 Jan 2026 00:00:00 +0000

Facial expression recognition develops rapidly during infancy and improves from childhood to adulthood. As a critical component of social communication, this skill enables individuals to interpret others’ emotions and intentions. However, the brain mechanisms driving the development of this skill remain largely unclear due to the difficulty of obtaining data with both high spatial and temporal resolution from young children. By analyzing intracranial EEG data collected from childhood (5–10 years old) and post-childhood groups (13–55 years old), we find differential involvement of high-level brain area in processing facial expression information. For the post-childhood group, both the posterior superior temporal cortex (pSTC) and the dorsolateral prefrontal cortex (DLPFC) encode facial emotion features from a high-dimensional space. However, in children, the facial expression information is only significantly represented in the pSTC, not in the DLPFC. Furthermore, the encoding of complex emotions in pSTC is shown to increase with age. Taken together, young children rely more on low-level sensory areas than on the prefrontal cortex for facial emotion processing, suggesting that the prefrontal cortex matures with development to enable a full understanding of facial emotions, especially complex emotions that require social and life experience to comprehend.

Computational modelling identifies key determinants of subregion-specific dopamine dynamics in the striatum

frejahh@sund.ku.dk (Aske Ejdrup) — Fri, 23 Jan 2026 00:00:00 +0000

Striatal dopamine (DA) release regulates reward-related learning and motivation and is believed to consist of a short-lived phasic and continuous tonic component. Here, we build a large-scale three-dimensional model of extracellular DA dynamics in dorsal (DS) and ventral striatum (VS). The model predicts rapid dynamics in DS with little to no basal DA and slower dynamics in the VS enabling build-up of tonic DA levels. These regional differences do not reflect release-related phenomena but rather differential dopamine transporter (DAT) activity. Interestingly, our simulations posit DAT nanoclustering as a possible regulator of this activity. Receptor binding simulations show that D1 receptor occupancy follows extracellular DA concentration with milliseconds delay, while D2 receptors do not respond to brief pauses in firing but rather integrate DA signal over seconds. Summarised, our model distills recent experimental observations into a computational framework that challenges prevailing paradigms of striatal DA signalling.

Anti-resonance in developmental signaling regulates cell fate decisions

M.S.Bauer@tudelft.nl (Erik Hopkins) — Wed, 21 Jan 2026 00:00:00 +0000

Cells process dynamic signaling inputs to regulate fate decisions during development. While oscillations or waves in key developmental pathways, such as Wnt, have been widely observed, the principles governing how cells decode these signals remain unclear. By leveraging optogenetic control of the Wnt signaling pathway in both HEK293T cells and H9 human embryonic stem cells, we systematically map the relationship between signal frequency and downstream pathway activation. We find that cells exhibit a minimal response to Wnt at certain frequencies, a behavior we term anti-resonance. We developed both detailed biochemical and simplified hidden variable models that explain how anti-resonance emerges from the interplay between fast and slow pathway dynamics. Remarkably, we find that frequency directly influences cell fate decisions involved in human gastrulation; signals delivered at anti-resonant frequencies result in dramatically reduced mesoderm differentiation. Our work reveals a previously unknown mechanism of how cells decode dynamic signals and how anti-resonance may filter against spurious activation. These findings establish new insights into how cells decode dynamic signals with implications for tissue engineering, regenerative medicine, and cancer biology.

Predicting human decision-making across task conditions via individuality transfer

higashi@comm.eng.osaka-u.ac.jp (Hiroshi Higashi) — Mon, 19 Jan 2026 00:00:00 +0000

Predicting an individual's behavior in one task condition based on their behavior in a different condition is a key challenge in modeling individual decision-making tendencies. We propose a novel framework that addresses this challenge by leveraging neural networks and introducing a concept we term the ‘individual latent representation’. This representation, extracted from behavior in a ‘source’ task condition via an encoder network, captures an individual's unique decision-making tendencies. A decoder network then utilizes this representation to generate the weights of a task-specific neural network (a ‘task solver’), which predicts the individual's behavior in a ‘target' task condition. We demonstrate the effectiveness of our approach in two distinct decision-making tasks: a value-guided task and a perceptual task. Our framework offers a robust and generalizable approach for parameterizing individual variability, providing a promising pathway toward computational modeling at the individual level—replicating individuals in silico.

Biologically informed cortical models predict optogenetic perturbations

guillaume.bellec@epfl.ch (Carl CH Petersen) — Fri, 16 Jan 2026 00:00:00 +0000

A recurrent neural network fitted to large electrophysiological datasets may help us understand the chain of cortical information transmission. In particular, successful network reconstruction methods should enable a model to predict the response to optogenetic perturbations. We test recurrent neural networks (RNNs) fitted to electrophysiological datasets on unseen optogenetic interventions and measure that generic RNNs used predominantly in the field generalize poorly on these perturbations. Our alternative RNN model adds biologically informed inductive biases like structured connectivity of excitatory and inhibitory neurons and spiking neuron dynamics. We measure that some biological inductive biases improve the model prediction on perturbed trials in a simulated dataset and a dataset recorded in mice in vivo. Furthermore, we show in theory and simulations that gradients of the fitted RNN can be used to target micro-perturbations in the recorded circuits and discuss the potential utility to bias an animal’s behavior and study cortical circuit mechanisms.

Targeted computational design of an interleukin-7 superkine with enhanced folding efficiency and immunotherapeutic efficacy

chu@ibms.sinica.edu.tw (Che-Ming Jack Hu) — Fri, 16 Jan 2026 00:00:00 +0000

Interleukin-7 (IL-7) plays a central role in maintaining T cell development and immune homeostasis, and enhancing the cytokine’s immune-stimulatory functionality has broad therapeutic implications against various oncological malignancies. Herein, we show a computationally designed IL7 superkine, Neo-7, which exhibits enhanced folding efficiency and superior binding affinity to its cognate receptors. To streamline the protein candidate prediction and validation process, the loop region of IL7 was strategically targeted for redesign while most of the receptor-interacting regions were preserved. Leveraging advanced computational tools such as AlphaFold2, we show loop remodeling to rectify structural irregularities that allow for iterative stabilization of protein backbone and lead to identification of beneficial mutations conducive to receptor engagement. Neo-7 superkine shows improved thermostability and production yield, and it exhibits heightened immune-stimulatory and anticancer effect in C57BL/6 J mice. Neo-7 addresses intrinsic developability limitations of IL-7, including inefficient folding, aggregation propensity, and suboptimal receptor engagement, while in vivo pharmacokinetic limitations of wild-type IL-7 were addressed separately through Fc fusion. These findings underscore the utility of a targeted computational approach for de novo cytokine development.

Nonlinear transcriptional responses to gradual modulation of transcription factor dosage

julia.domingo.espinos@gmail.com (John A Morris) — Wed, 14 Jan 2026 00:00:00 +0000

Genomic loci associated with common traits and diseases are typically non-coding and likely impact gene expression, sometimes coinciding with rare loss-of-function variants in the target gene. However, our understanding of how gradual changes in gene dosage affect molecular, cellular, and organismal traits is currently limited. To address this gap, we induced gradual changes in gene expression of four genes using CRISPR activation and inactivation in human-derived K562 cells. Downstream transcriptional consequences of dosage modulation of three master trans-regulators associated with blood cell traits (GFI1B, NFE2, and MYB) were examined using targeted single-cell multimodal sequencing. We showed that guide tiling around the TSS is the most effective way to modulate cis gene expression across a wide range of fold changes, with further effects from chromatin accessibility and histone marks that differ between the inhibition and activation systems. Our single-cell data allowed us to precisely detect subtle to large gene expression changes in dozens of trans genes, revealing that many responses to dosage changes of these three TFs are nonlinear, including non-monotonic behaviours, even when constraining the fold changes of the master regulators to a copy number gain or loss. We found that the dosage properties are linked to gene constraint and that some of these nonlinear responses are enriched for disease and GWAS genes. Overall, our study provides a straightforward and scalable method to precisely modulate gene expression and gain insights into its downstream consequences at high resolution.

Predicting the effect of CRISPR-Cas9-based epigenome editing

isaac.hilton@rice.edu (Alan Cabrera) — Mon, 12 Jan 2026 00:00:00 +0000

Epigenetic regulation orchestrates mammalian transcription, but functional links between them remain elusive. To tackle this problem, we use epigenomic and transcriptomic data from 13 ENCODE cell types to train machine learning models to predict gene expression from histone post-translational modifications (PTMs), achieving transcriptome-wide correlations of ∼0.70−0.79 for most cell types. Our models recapitulate known associations between histone PTMs and expression patterns, including predicting that acetylation of histone subunit H3 lysine residue 27 (H3K27ac) near the transcription start site (TSS) significantly increases expression levels. To validate this prediction experimentally and investigate how natural vs. engineered deposition of H3K27ac might differentially affect expression, we apply the synthetic dCas9-p300 histone acetyltransferase system to 8 genes in the HEK293T cell line and to 5 genes in the K562 cell line. Further, to facilitate model building, we perform MNase-seq to map genome-wide nucleosome occupancy levels in HEK293T. We observe that our models perform well in accurately ranking relative fold-changes among genes in response to the dCas9-p300 system; however, their ability to rank fold-changes within individual genes is noticeably diminished compared to predicting expression across cell types from their native epigenetic signatures. Our findings highlight the need for more comprehensive genome-scale epigenome editing datasets, better understanding of the actual modifications made by epigenome editing tools, and improved causal models that transfer better from endogenous cellular measurements to perturbation experiments. Together, these improvements would facilitate the ability to understand and predictably control the dynamic human epigenome with consequences for human health.

Reconstructing voice identity from noninvasive auditory cortex recordings

charlylmth@gmail.com (Bruno L Giordano) — Thu, 08 Jan 2026 00:00:00 +0000

The cerebral processing of voice information is known to engage, in human as well as non-human primates, ‘temporal voice areas’ (TVAs) that respond preferentially to conspecific vocalizations. However, how voice information is represented by neuronal populations in these areas, particularly speaker identity information, remains poorly understood. Here, we used a deep neural network (DNN) to generate a high-level, small-dimension representational space for voice identity—the ‘voice latent space’ (VLS)—and examined its linear relation with cerebral activity via encoding, representational similarity, and decoding analyses. We find that the VLS maps onto fMRI measures of cerebral activity in response to tens of thousands of voice stimuli from hundreds of different speaker identities and better accounts for the representational geometry for speaker identity in the TVAs than in A1. Moreover, the VLS allowed TVA-based reconstructions of voice stimuli that preserved essential aspects of speaker identity as assessed by both machine classifiers and human listeners. These results indicate that the DNN-derived VLS provides high-level representations of voice identity information in the TVAs.

Atypical collective oscillatory activity in cardiac tissue uncovered by optogenetics

a.teplenin@lumc.nl (Alexander S Teplenin) — Wed, 07 Jan 2026 00:00:00 +0000

Many biological processes emerge as frequency-dependent responses to trains of external stimuli. Heart rhythm disturbances, that is cardiac arrhythmias, are important examples as they are often triggered by specific patterns of preceding stimuli. In this study, we investigated how ectopic arrhythmias can be induced by external stimuli in cardiac tissue containing a localised area of depolarisation. Using optogenetic in vitro experiments and in silico modelling, we systematically explored the dynamics of these arrhythmias, which are characterised by local oscillatory activity, by gradually altering the degree of depolarisation in a predefined region. Our findings reveal a bi-stable system, in which transitions between oscillatory ectopic activity and a quiescent state can be precisely controlled, that is by adjusting the number and frequency of propagating waves through the depolarised area oscillations could be turned on or off. These frequency-dependent responses arise from collective mechanisms involving stable, non-self-oscillatory cells, contrasting with the typical role of self-oscillations in individual units within biophysical systems. To further generalise these findings, we demonstrated similar frequency selectivity and bi-stability in a simplified reaction–diffusion model. This suggests that complex ionic cell dynamics are not required to reproduce these effects; rather, simpler non-linear systems can replicate similar behaviour, potentially extending beyond the cardiac context.

Morphogenesis and morphometry of brain folding patterns across species

lmahadev@g.harvard.edu (Chunzi Liu) — Mon, 29 Dec 2025 00:00:00 +0000

Evolutionary adaptations associated with the formation of a folded cortex in many mammalian brains are thought to be a critical specialization associated with higher cognitive function. The dramatic surface expansion and highly convoluted folding of the cortex during early development is a theme with variations that suggest the need for a comparative study of cortical gyrification. Here, we use a combination of physical experiments using gels, computational morphogenesis, and geometric morphometrics to study the folding of brains across different species. Starting with magnetic resonance images of brains of a newborn ferret, a fetal macaque, and a fetal human, we construct two-layer physical gel brain models that swell superficially in a solvent, leading to folding patterns similar to those seen in vivo. We then adopt a three-dimensional continuum model based on differential growth to simulate cortical folding in silico. Finally, we deploy a comparative morphometric analysis of the in vivo, in vitro, and in silico surface buckling patterns across species. Our study shows that a simple mechanical instability driven by differential growth suffices to explain cortical folding and suggests that variations in the tangential growth and different initial geometries are sufficient to explain the differences in cortical folding across species.

Predictive modeling of hematoma expansion from non-contrast computed tomography in spontaneous intracerebral hemorrhage patients

ni8vb@uvahealth.org (Andrea Becceril-Gaitan) — Tue, 23 Dec 2025 00:00:00 +0000

Hematoma expansion is a consistent predictor of poor neurological outcome and mortality after spontaneous intracerebral hemorrhage (ICH). An incomplete understanding of its biophysiology has limited early preventative intervention. Transport-based morphometry (TBM) is a mathematical modeling technique that uses a physically meaningful metric to quantify and visualize discriminating image features that are not readily perceptible to the human eye. We hypothesized that TBM could discover relationships between hematoma morphology on initial Non-Contrast Computed Tomography (NCCT) and hematoma expansion. 170 spontaneous ICH patients enrolled in the multi-center international Virtual International Trials of Stroke Archive (VISTA-ICH) with time-series NCCT data were used for model derivation. Its performance was assessed on a test dataset of 170 patients from the Ethnic/Racial Variations of Intracerebral Hemorrhage (ERICH) study. A unique transport-based representation was produced from each presentation NCCT hematoma image to identify morphological features of expansion. The principal hematoma features identified by TBM were larger size, density heterogeneity, shape irregularity, and peripheral density distribution. These were consistent with clinician-identified features of hematoma expansion, corroborating the hypothesis that morphological characteristics of the hematoma promote future growth. Incorporating these traits into a multivariable model comprising morphological, spatial, and clinical information achieved an AUROC of 0.71 for quantifying 24 hr hematoma expansion risk in the test dataset. This outperformed existing clinician protocols and alternate machine learning methods, suggesting that TBM detected features with improved precision than by visual inspection alone. This pre-clinical study presents a quantitative and interpretable method for discovery and visualization of NCCT biomarkers of hematoma expansion in ICH patients. Because TBM has a direct physical meaning, its modeling of NCCT hematoma features can inform hypotheses for hematoma expansion mechanisms. It has potential future application as a clinical risk stratification tool.

Human EEG and artificial neural networks reveal disentangled representations and processing timelines of object real-world size and depth in natural images

zitonglu@mit.edu (Julie Golomb) — Mon, 22 Dec 2025 00:00:00 +0000

Remarkably, human brains have the ability to accurately perceive and process the real-world size of objects, despite vast differences in distance and perspective. While previous studies have delved into this phenomenon, distinguishing the processing of real-world size from other visual properties, like depth, has been challenging. Using the THINGS EEG2 dataset with human EEG recordings and more ecologically valid naturalistic stimuli, our study combines human EEG and representational similarity analysis to disentangle neural representations of object real-world size from retinal size and perceived depth, leveraging recent datasets and modeling approaches to address challenges not fully resolved in previous work. We report a representational timeline of visual object processing: object real-world depth processed first, then retinal size, and finally, real-world size. Additionally, we input both these naturalistic images and object-only images without natural background into artificial neural networks. Consistent with the human EEG findings, we also successfully disentangled representation of object real-world size from retinal size and real-world depth in all three types of artificial neural networks (visual-only ResNet, visual-language CLIP, and language-only Word2Vec). Moreover, our multi-modal representational comparison framework across human EEG and artificial neural networks reveals real-world size as a stable and higher-level dimension in object space incorporating both visual and semantic information. Our research provides a temporally resolved characterization of how certain key object properties – such as object real-world size, depth, and retinal size – are represented in the brain, which offers further advances and insights into our understanding of object space and the construction of more brain-like visual models.

Exploiting fluctuations in gene expression to detect causal interactions between genes

andreas.hilfinger@utoronto.ca (Andreas Hilfinger) — Tue, 16 Dec 2025 00:00:00 +0000

Characterizing and manipulating cellular behavior requires a mechanistic understanding of the causal interactions between cellular components. We present an approach to detect causal interactions between genes without the need to perturb the physiological state of cells. This approach exploits naturally occurring cell-to-cell variability which is experimentally accessible from static population snapshots of genetically identical cells without the need to follow cells over time. Our main contribution is a simple mathematical relation that constrains the propagation of gene expression noise through biochemical reaction networks. This relation allows us to rigorously interpret fluctuation data even when only a small part of a complex gene regulatory process can be observed. We show how this relation can, in theory, be exploited to detect causal interactions by synthetically engineering a passive reporter of gene expression, akin to the established ‘dual reporter assay’. While the focus of our contribution is theoretical, we also present an experimental proof-of-principle to demonstrate the real-world applicability of our approach in certain circumstances. Our experimental data suggest that the method can detect causal interactions in specific synthetic gene regulatory circuits in Escherichia coli, confirming our theoretical result in a narrow set of controlled experimental settings. Further work is needed to show that the approach is practical on a large scale, with naturally occurring gene regulatory networks, or in organisms other than E. coli.

A dynamic scale-mixture model of motion in natural scenes

sepalmer@uchicago.edu (Jared M Salisbury) — Tue, 09 Dec 2025 00:00:00 +0000

Some of the most important tasks of visual and motor systems involve estimating the motion of objects and tracking them over time. Such systems evolved to meet the behavioral needs of the organism in its natural environment and may therefore be adapted to the statistics of motion it is likely to encounter. By tracking the movement of individual points in movies of natural scenes, we begin to identify common properties of natural motion across scenes. As expected, objects in natural scenes move in a persistent fashion, with velocity correlations lasting hundreds of milliseconds. More subtly, but crucially, we find that the observed velocity distributions are heavy-tailed and can be modeled as a Gaussian scale-mixture. Extending this model to the time domain leads to a dynamic scale-mixture model, consisting of a Gaussian process multiplied by a positive scalar quantity with its own independent dynamics. Dynamic scaling of velocity arises naturally as a consequence of changes in object distance from the observer and may approximate the effects of changes in other parameters governing the motion in a given scene. This modeling and estimation framework has implications for the neurobiology of sensory and motor systems, which need to cope with these fluctuations in scale in order to represent motion efficiently and drive fast and accurate tracking behavior.

The electrogenicity of the Na⁺/K⁺-ATPase poses challenges for computation in highly active spiking cells

s.schreiber@hu-berlin.de (Jan-Hendrik Schleimer) — Wed, 03 Dec 2025 00:00:00 +0000

The evolution of the Na⁺/K⁺-ATPase laid the foundation for ion homeostasis and electrical signaling. While not required for restoration of ionic gradients, the electrogenicity of the pump (resulting from its 3:2 stoichiometry) is useful to prevent runaway activity. As we show here, electrogenicity could also come with disadvantageous side effects: (1) an activity-dependent shift in a cell’s baseline firing and (2) interference with computation, disturbing network entrainment when inputs change strongly. We exemplify these generic effects in a mathematical model of the weakly electric fish electrocyte, which spikes at hundreds of Hz and is exposed to abrupt rate changes when producing behaviorally relevant communication signals. We discuss biophysical strategies that may allow cells to mitigate the consequences of electrogenicity at additional metabolic cost and postulate an interesting role for a voltage dependence of the Na⁺/K⁺-ATPase. Our work shows that the pump’s electrogenicity can open an additional axis of vulnerability that may play a role in brain disease.