arXiv Quantitative Biology @arxiv_bio@qoto.org

Predicting protein complex structures is essential for protein function analysis, protein design, and drug discovery. While AI methods like AlphaFold can predict accurate structural models for many protein complexes, reliably estimating the quality of these predicted models (estimation of model accuracy, or EMA) for model ranking and selection remains a major challenge. A key barrier to developing effective machine learning-based EMA methods is the lack of large, diverse, and well-annotated datasets for training and evaluation. To address this gap, we introduce PSBench, a benchmark suite comprising four large-scale, labeled datasets generated during the 15th and 16th community-wide Critical Assessment of Protein Structure Prediction (CASP15 and CASP16). PSBench includes over one million structural models covering a wide range of protein sequence lengths, complex stoichiometries, functional classes, and modeling difficulties. Each model is annotated with multiple complementary quality scores at the global, local, and interface levels. PSBench also provides multiple evaluation metrics and baseline EMA methods to facilitate rigorous comparisons. To demonstrate PSBench's utility, we trained and evaluated GATE, a graph transformer-based EMA method, on the CASP15 data. GATE was blindly tested in CASP16 (2024), where it ranked among the top-performing EMA methods. These results highlight PSBench as a valuable resource for advancing EMA research in protein complex modeling. PSBench is publicly available at: https://github.com/BioinfoMachineLearning/PSBench.

The holonomic brain theory, originally formulated to account for the need of non-local memory encoding in cognitive systems, could gain new theoretical traction when integrated with holographic principles from physics, most notably the AdS/CFT correspondence. Recent findings in neuroscience suggest that conformal field theories (CFTs), emerging at critical points across spatiotemporal scales in neural dynamics, are essential for brain function. Concurrently, black-brane geometries, long studied in gravitational physics, can find unexpected analogues in the interplay of active matter dynamics and the brain s neuroanatomical organization. Motivated by these parallels, we posit a generalized holographic framework and interrogate its validity through the fluid/gravity duality; a correspondence linking hydrodynamic equations to gravitational spacetime metrics. In this work, we explore the holographic principles at the Navier-Stokes regime, demonstrating that holography can model key neurophysiological mechanisms: cerebral autoregulation (the brain s hemodynamic self-stabilization) and neurovascular coupling (the dynamic neuron-bloodflow interplay). This work bridges holography, active matter physics, and neuroscience, proposing a unified framework to decode the brain s multiscale organization, its resilience to perturbations, and its computational capabilities. By grounding neurovascular physiology in gravitational duals, we open pathways to reinterpret brain function through the lens of emergent spacetime geometry.

Brain network analysis plays a crucial role in diagnosing and monitoring neurodegenerative disorders such as Alzheimer's disease (AD). Existing approaches for constructing structural brain networks from diffusion tensor imaging (DTI) often rely on specialized toolkits that suffer from inherent limitations: operator subjectivity, labor-intensive workflows, and restricted capacity to capture complex topological features and disease-specific biomarkers. To overcome these challenges and advance computational neuroimaging instrumentation, ConnectomeDiffuser is proposed as a novel diffusion-based framework for automated end-to-end brain network construction from DTI. The proposed model combines three key components: (1) a Template Network that extracts topological features from 3D DTI scans using Riemannian geometric principles, (2) a diffusion model that generates comprehensive brain networks with enhanced topological fidelity, and (3) a Graph Convolutional Network classifier that incorporates disease-specific markers to improve diagnostic accuracy. ConnectomeDiffuser demonstrates superior performance by capturing a broader range of structural connectivity and pathology-related information, enabling more sensitive analysis of individual variations in brain networks. Experimental validation on datasets representing two distinct neurodegenerative conditions demonstrates significant performance improvements over other brain network methods. This work contributes to the advancement of instrumentation in the context of neurological disorders, providing clinicians and researchers with a robust, generalizable measurement framework that facilitates more accurate diagnosis, deeper mechanistic understanding, and improved therapeutic monitoring of neurodegenerative diseases such as AD.

The COVID-19 pandemic response relied heavily on statistical and machine learning models to predict key outcomes such as case prevalence and fatality rates. These predictions were instrumental in enabling timely public health interventions that helped break transmission cycles. While most existing models are grounded in traditional epidemiological data, the potential of alternative datasets, such as those derived from genomic information and human behavior, remains underexplored. In the current study, we investigated the usefulness of diverse modalities of feature sets in predicting case surges. Our results highlight the relative effectiveness of biological (e.g., mutations), public health (e.g., case counts, policy interventions) and human behavioral features (e.g., mobility and social media conversations) in predicting country-level case surges. Importantly, we uncover considerable heterogeneity in predictive performance across countries and feature modalities, suggesting that surge prediction models may need to be tailored to specific national contexts and pandemic phases. Overall, our work highlights the value of integrating alternative data sources into existing disease surveillance frameworks to enhance the prediction of pandemic dynamics.

Bi-directional brain computer interfaces (BD-BCIs) may restore brain-controlled walking and artificial leg sensation after spinal cord injury. Current BD-BCIs provide only simplistic "tingling" feedback, which lacks proprioceptive information to perceive critical gait events (leg swing, double support). This information must also be perceived adequately fast to facilitate timely motor responses. Here, we investigated utilizing primary sensory cortex (S1) direct cortical electrical stimulation (DCES) to deliver leg proprioceptive information and measured response times to artificial leg sensations. Subjects with subdural electrocorticogram electrodes over S1 leg areas participated in two tasks: (1) Proprioceptive acuity: subjects identified the difference between DCES-induced percepts emulating various leg swing speeds; (2) Sensory response: measuring subjects' reaction time to DCES-induced leg sensations, with DCES-hand, visual and auditory control conditions. Three subjects were recruited. Only one completed the proprioceptive assessment, achieving 80%, 70%, 60%, and 53% accuracy in discriminating between fast/slow, fast/medium, medium/slow, and same speeds, respectively (p-value=1.9x10$^{-5}$). Response times for leg/hand percepts were 1007$\pm$413/599$\pm$171 ms, visual leg/hand responses were 528$\pm$137/384$\pm$84 ms, and auditory leg/hand responses were 393$\pm$106/352$\pm$93 ms, respectively. These results suggest proprioceptive information can be delivered artificially, but perception may be significantly delayed. Future work should address improving acuity, reducing response times, and expanding sensory modalities.

Attractor dynamics are a hallmark of many complex systems, including the brain. Understanding how such self-organizing dynamics emerge from first principles is crucial for advancing our understanding of neuronal computations and the design of artificial intelligence systems. Here we formalize how attractor networks emerge from the free energy principle applied to a universal partitioning of random dynamical systems. Our approach obviates the need for explicitly imposed learning and inference rules and identifies emergent, but efficient and biologically plausible inference and learning dynamics for such self-organizing systems. These result in a collective, multi-level Bayesian active inference process. Attractors on the free energy landscape encode prior beliefs; inference integrates sensory data into posterior beliefs; and learning fine-tunes couplings to minimize long-term surprise. Analytically and via simulations, we establish that the proposed networks favor approximately orthogonalized attractor representations, a consequence of simultaneously optimizing predictive accuracy and model complexity. These attractors efficiently span the input subspace, enhancing generalization and the mutual information between hidden causes and observable effects. Furthermore, while random data presentation leads to symmetric and sparse couplings, sequential data fosters asymmetric couplings and non-equilibrium steady-state dynamics, offering a natural extension to conventional Boltzmann Machines. Our findings offer a unifying theory of self-organizing attractor networks, providing novel insights for AI and neuroscience.

Understanding how protein mutations affect protein-nucleic acid binding is critical for unraveling disease mechanisms and advancing therapies. Current experimental approaches are laborious, and computational methods remain limited in accuracy. To address this challenge, we propose a novel topological machine learning model (TopoML) combining persistent Laplacian (from topological data analysis) with multi-perspective features: physicochemical properties, topological structures, and protein Transformer-derived sequence embeddings. This integrative framework captures robust representations of protein-nucleic acid binding interactions. To validate the proposed method, we employ two datasets, a protein-DNA dataset with 596 single-point amino acid mutations, and a protein-RNA dataset with 710 single-point amino acid mutations. We show that the proposed TopoML model outperforms state-of-the-art methods in predicting mutation-induced binding affinity changes for protein-DNA and protein-RNA complexes.

Neural communication across the cortex, subcortex, and cerebellum is orchestrated by the structural connectome, forming the indispensable anatomical framework for capabilities spanning from elementary motor actions to higher cognitive functions. Yet, despite this importance, the core organizational rules that govern this connectivity remain insufficiently understood. Here we show, for the first time, how the integrated cortical, subcortical, and cerebellar brain areas shape the structural architecture of the whole brain. We find dense structural clusters, which differ in composition and arrangement, vertically transverse the canonical cortical, subcortical, and cerebellar boundaries. These clusters are centralized by a global rich club of predominantly subcortical, alongside cortical hub regions. Congruently, we find that subcortical hubs are not only the most widely connected brain areas but are also leading overall structural integration. Nearly all larger subcortical structures encompass these hub regions, but they also exhibit brain regions with fewer but more specialized connections, pointing toward functional heterogeneity in these structures themselves. Our findings move beyond traditional cortico-centric analysis, offering an initial and global perspective for understanding overall structural connectivity.

In recent years, there is a growing need and opportunity to use online platforms for psychophysics research. Online experiments make it possible to evaluate large and diverse populations remotely and quickly, complementing laboratory-based research. However, developing and running online psychophysics experiments poses several challenges: i) a high barrier-to-entry for researchers who often need to learn complex code-based platforms, ii) an uncontrolled experimental environment, and iii) questionable credibility of the participants. Here, we introduce an open-source Modular Online Psychophysics Platform (MOPP) to address these challenges. Through the simple web-based interface of MOPP, researchers can build modular experiments, share them with others, and copy or modify tasks from each others environments. MOPP provides built-in features to calibrate for viewing distance and to measure visual acuity. It also includes email-based and IP-based authentication, and reCAPTCHA verification. We developed five example psychophysics tasks, that come preloaded in the environment, and ran a pilot experiment which was hosted on the AWS (Amazon Web Services) cloud. Pilot data collected for these tasks yielded similar results to those reported in laboratory settings. MOPP can thus help researchers collect large psychophysics datasets online, with reduced turnaround time, and in a standardized manner.

Advances in AI, particularly LLMs, have dramatically shortened drug discovery cycles by up to 40% and improved molecular target identification. However, these innovations also raise dual-use concerns by enabling the design of toxic compounds. Prompting Moremi Bio Agent without the safety guardrails to specifically design novel toxic substances, our study generated 1020 novel toxic proteins and 5,000 toxic small molecules. In-depth computational toxicity assessments revealed that all the proteins scored high in toxicity, with several closely matching known toxins such as ricin, diphtheria toxin, and disintegrin-based snake venom proteins. Some of these novel agents showed similarities with other several known toxic agents including disintegrin eristostatin, metalloproteinase, disintegrin triflavin, snake venom metalloproteinase, corynebacterium ulcerans toxin. Through quantitative risk assessments and scenario analyses, we identify dual-use capabilities in current LLM-enabled biodesign pipelines and propose multi-layered mitigation strategies. The findings from this toxicity assessment challenge claims that large language models (LLMs) are incapable of designing bioweapons. This reinforces concerns about the potential misuse of LLMs in biodesign, posing a significant threat to research and development (R&D). The accessibility of such technology to individuals with limited technical expertise raises serious biosecurity risks. Our findings underscore the critical need for robust governance and technical safeguards to balance rapid biotechnological innovation with biosecurity imperatives.

Natural protein sequences somehow encode the structural forms that these molecules adopt. Recent developments in structure-prediction are agnostic to the mechanisms by which proteins fold and represent them as static objects. However, the amino acid sequences also encode information about how the folding process can happen, and how variations in the sequences impact on the populations of the distinct structural forms that proteins acquire. Here we present a method to infer protein folding dynamics based only on sequence information. For this, we will rely first on the obtention of a precise 'evolutionary field' from the observed variations in the sequences of homologous proteins. We then show how to map the energetics to a coarse-grained folding model where the protein is treated as a string of foldons that interact. We then describe how, for any given protein sequence of a family, the equilibrium folding curve can be computed and how the emergence of protein folding sub-domains can be identified. We finally present protocols to analyze how mutations perturb both the folding stability and the cooperativity, that represent predictions for a deep-mutational scan of a protein of interest.

To be fully useful for public health practice, models for epidemic response must be able to do more than predict -- it is also important to incorporate the mechanisms underlying transmission dynamics to enable policymakers and practitioners to be able to evaluate what-if scenarios and intervention options. However, most mechanistic models suffer from uncertainty in both the parameters (e.g., parameter unidentifiability) and the model structure itself, which can hinder both successful parameter estimation and model interpretation. To enable rapid development of interpretable and parsimonious mechanistic models, we use penalized regression and covariate selection methods to integrate parameter identifiability and model selection directly into the parameter estimation procedure for (in this case) traditional ordinary differential equation (ODE) models. For both simulated and real-world epidemiological data, we demonstrate that the LASSO-ODE framework is highly effective in selecting a parsimonious, identifiable model from larger, more realistic but potentially unidentifiable models, from realistically sparse data with only a single measured compartment and multiple latent (unobserved) variables. While we focus on epidemic models in this paper as a case study, these same approaches are applicable to a wide range of application areas that are faced with relatively sparse data but a need for realistic mechanistic models (e.g. mathematical oncology and mathematical biology more broadly). Additionally, the cross-validation techniques designed for time series data introduced in our study can be used across a range of time series analysis and modeling approaches.

We consider a metapopulation model of cholera describing explicitly the movement of individuals and contaminated water between locations as well as a simple vaccination mechanism. The global stability of the disease-free equilibrium point when the location-specific reproduction numbers are all less than unity is established. We then conduct some numerical investigation of the model.

A major goal of neuroscience is to understand brain computations during visual processing in naturalistic settings. A dominant approach is to use image-computable deep neural networks trained with different task objectives as a basis for linear encoding models. However, in addition to requiring tuning a large number of parameters, the linear encoding approach ignores the structure of the feature maps both in the brain and the models. Recently proposed alternatives have focused on decomposing the linear mapping to spatial and feature components but focus on finding static receptive fields for units that are applicable only in early visual areas. In this work, we employ the attention mechanism used in the transformer architecture to study how retinotopic visual features can be dynamically routed to category-selective areas in high-level visual processing. We show that this computational motif is significantly more powerful than alternative methods in predicting brain activity during natural scene viewing, across different feature basis models and modalities. We also show that this approach is inherently more interpretable, without the need to create importance maps, by interpreting the attention routing signal for different high-level categorical areas. Our approach proposes a mechanistic model of how visual information from retinotopic maps can be routed based on the relevance of the input content to different category-selective regions.

Background In proteomics, the most probable localizations of post-translational modifications are assessed by localization scores evaluating the likelihood of a given modification to occupy a site on a peptide sequence. When identifying highly modified peptides, localization scores for different modifications can return conflicting results, stacking modifications on the same amino acid. Here, we propose a graph-based approach that assigns modifications to sites in a way that maximizes localization scores while avoiding conflicting assignments. Results The algorithm is implemented as both a standalone Python program and in the compomics-utilities Java library. Our graph-based approach showed the ability to match complex combinations of modifications and acceptor sites, allowing the processing of thousands of peptides in a few seconds. Conclusions Our graph-based approach to modification site assignment allows distributing multiple modifications in a way that maximizes individual localization scores. Having an optimal modification site assignment is important for spectrum annotation and biological interpretation.

Interspecific communication plays a critical role in mediating human-animal interactions, particularly in contexts involving access to anthropogenic resources. This study investigates the influence of human gazing on the begging strategies of free-ranging dogs in urban and peri-urban environments. Begging behaviour, commonly observed in dogs seeking food from humans, offers insights into their behavioural flexibility and cognitive attunement to human social cues. We observed 650 adult dogs in both solitary and group settings to assess how social context shapes the expression of begging behaviour in free-ranging dogs. Our findings indicate that solitary dogs beg more frequently than those in groups, and that female dogs exhibit higher rates of begging, predominantly through passive strategies. Moreover, dogs modulate their active begging in response to subtle variations in human gazing and food availability. These results suggest that passive begging is influenced by stable situational factors such as sex and group composition, while active begging is more responsive to immediate environmental cues, including human attentional state. Collectively, our findings highlight the social competence and behavioural plasticity of free-ranging dogs in navigating interspecies interactions, and contribute to a broader understanding of how communicative strategies evolve in response to social and ecological pressures.

Designing metal-organic frameworks (MOFs) with novel chemistries is a long-standing challenge due to their large combinatorial space and the complex 3D arrangements of building blocks. While recent deep generative models have enabled scalable MOF generation, they assume (1) a fixed set of building blocks and (2) known ground-truth local block-wise 3D coordinates. However, this limits their ability to (1) design novel MOFs and (2) generate the structure using novel building blocks. We propose a two-stage de novo MOF generation framework that overcomes these limitations by modeling both chemical and geometric degrees of freedom. First, we train a SMILES-based autoregressive model to generate novel metal and organic building blocks, paired with cheminformatics for 3D structure initialization. Second, we introduce a flow-matching model that predicts translations, rotations, and torsional angles to assemble flexible blocks into valid 3D frameworks. Our experiments demonstrate improved reconstruction accuracy, the generation of valid, novel, and unique MOFs, and the ability of our model to create novel building blocks.

We introduce the SoftBart approach from Bayesian ensemble learning to estimate the relationship between multipollutant mixtures and health on chronic exposures in epidemiology research. This approach offers several key advantages over existing methods: (1) it is computationally efficient and well-suited for analyzing large datasets; (2) it is flexible in estimating various correlated nonlinear functions simultaneously; and (3) it accurately identifies active variables within highly correlated multipollutant mixtures. Through simulations, we demonstrate the method's superiority by comparing its accuracy in estimating and quantifying uncertainties for both main and interaction effects with the commonly used method, BKMR. Last, we apply the method to analyze a multipollutant dataset with 10,110 participates from the Nurses' Health Study.

Aperiodic neural activity has been the subject of intense research interest lately as it could reflect on the cortical excitation/inhibition ratio, which is suspected to be affected in numerous clinical conditions. This phenomenon is characterized via the aperiodic scaling exponent $β$, equal to the spectral slope following log-log transformation of power spectra. Despite recent progress, however, most current methods do not take into consideration the plausible multimodal nature in the power spectra of neurophysiological recordings - i.e., $β$ might be different in low- ($β_{lo}$) and high-frequency ($β_{hi}$) regimes -, especially in case of $|β_{lo}|>|β_{hi}|$. Here we propose an algorithm, the multi-modal spectral parametrization method (MMSPM) that aims to account for this issue. MMSPM estimates $β_{lo}$ and $β_{hi}$ separately using a constrained, piece-wise regression technique, and also assesses if they are significantly different or instead the spectrum is indeed unimodal and can be characterized simply with broadband $β$. Here we present the MMSPM algorithm and evaluate its performance in silico on simulated power spectra. Then, we use MMSPM on resting-state electroencephalography (EEG) data collected from 19 young, healthy volunteers, as well as on a separate dataset of EEG recordings from 30 schizophrenia patients and 31 healthy controls, and demonstrate that broadband (0.1-100 Hz and 0.5-45 Hz) EEG spectra can indeed present a bimodality pattern with significantly steeper low-range ($<\sim2$ Hz) and flatter high-range scaling regimes (i.e., $|β_{lo}|>|β_{hi}|$). Clinical relevance: The MMSPM method characterizes aperiodic neural activity in distinct scaling regimes, which can be relevant in numerous pathological conditions such as dementia or schizophrenia.