Large language models (LLMs) have become increasingly useful computational models of human language processing, but it remains unclear whether vision-language learning makes text representations more human-like during natural reading. Here, we address this question by comparing tightly matched LLM and vision-language model (VLM) pairs under a strictly text-only setting, allowing us to isolate the effect of multimodal training history from online visual input or cross-modal fusion. We evaluate model alignment with a human natural-reading dataset that includes whole-cortex fMRI responses and synchronized eye-tracking saccades. Our findings demonstrate that multimodal pretraining may not confer a uniform, global advantage in human alignment during natural reading, indicating that language-internal representations remain the key factor for modeling human text processing. However, the VLM advantage could emerge more selectively when sentences contain stronger visual semantic content, with converging evidence from both fMRI and eye-movement alignments. Together, our findings provide a controlled in silico framework for testing how visual learning history shapes model-human alignment of language processing, suggesting that multimodal pretraining contributes selectively rather than globally to human-like language representations during natural reading.

Tabular data in knowledge-rich domains often carries a latent prior in the form of Boolean implication relationships (BIRs) between pairs of features. We mine such relationships with a sparse-exception binomial test. The mined implications form a typed directed graph, equivalent to a propositional rule base of 2-literal clauses. We encode this graph as the connectivity of a layered neural network, called BIRDNet, in which each hidden unit corresponds to one mined rule and binds only to its two features. We show two consequences of this design: First, the architecture is sparse by construction: at most $2/d$ of the weights in each BIR layer are active, where $d$ is the input dimension. Second, the model is interpretable: every trained unit keeps a stable symbolic identity, so rules can be read off the network without surrogate models. Unlike most neurosymbolic models, BIRDNet does not consume an external rule base; its structural prior is mined from the data. We evaluate BIRDNet on six transcriptomic and proteomic benchmarks. Our results show that BIRDNet stays within 0.02 AUROC of the strongest dense baseline, at a small accuracy cost, while using up to $96\times$ fewer active parameters than an architecture-matched dense MLP. First-layer rules recover known biological signatures across multiple cancer subtypes and tissue types, including canonical amplicons, lineage-defining co-expression modules, and immune-infiltration markers. Data and code are available at: https://github.com/MAHI-Group/BIRDNet.

Backpropagation is the core learning mechanism underlying deep learning. However, whether and how this algorithm is implemented in the brain remains highly debated. In particular, while forward activations of pretrained models reliably map onto the cortical hierarchy of visual processing, it is unknown whether backpropagated gradients exhibit a similar correspondence. Here, we address this question using functional magnetic resonance imaging (fMRI) and magnetoencephalography (MEG) recordings of human brain responses to natural images. For this, we extend standard encoding analyses of forward activations to map backpropagated gradients onto neural data. Focusing on a recent self-supervised vision model (DINOv3) and reproducing results on eight vision models, we find that backpropagated gradients can reliably predict both fMRI and MEG signals, specifically in higher-level visual cortex and for later latencies. However, the spatial and temporal organization of these backpropagated gradients in the brain diverges from the patterns expected under a biologically plausible backpropagation mechanism: specifically, both the order in which gradients are computed and their spatial organization diverge from the temporal and spatial hierarchies of the human brain. Together, these results suggest that, although deep networks and the brain may share similar representational content, they likely rely on fundamentally different mechanisms to learn those representations.

From molecular, cellular, to ecological systems, the modeling of biological processes often stands on the assumption that fast components immediately reach the equilibrium at each moment (quasi-steady state) and only slow components govern the relevant system dynamics. This quasi-steady state approximation (QSSA) simplifies the modeling but discards the effects of the relaxation towards each quasi-steady state. Unclear is the QSSA's suitability around the transition point, a specific condition where the system changes to a qualitatively different state. In this regard, we here derived a theoretical framework for the near-transition dynamics of biological systems, explicitly considering the relaxation processes overlooked by the QSSA. Numerical simulations verify our predictions for cellular decision-making, metabolic oscillations, and ecological cycles. Despite the extreme slowdown near the transition point, the QSSA alone misestimates the duration of the transition from one state to another. Moreover, the QSSA erroneously predicts the transition point itself for the onset of oscillations, while the relaxation dynamics facilitates or suppresses the oscillation onset with a counterintuitive time-delay effect. Common feedback interactions between biological components are pivotal to those relaxation effects. Our study provides an analytical foundation to understand the rich transient or rhythmic dynamics of interacting biological components near the transitions.

Phase separation of various materials has been studied for one and a half centuries. In the last two decades, phase separation of proteins and nucleic acids has received enormous attention, due its relevance to cellular functions. However, many of the observations on the resulting biomolecular condensates lack a theoretical underpinning. The first goal of this Account is to put forward theoretical frameworks for the phase-separation propensities, material states, and material properties of biomolecular condensates. Using these frameworks, I rationalize mechanistic interpretations from our recent experimental and computational studies, and synthesize these studies with prior literature to draw new conclusions. For phase-separation propensities, I relate the threshold (or saturation) concentration to the excess chemical potential in the dense phase, which in turn depends on intermolecular interaction strength and valency. For material states, I posit that liquid droplets form via complete phase separation, whereas amorphous dense liquids, reversible aggregates, and gels arise from premature termination of spinodal decomposition, due to overly weak or overly strong interactions or directional interactions. In particular, gels and aggregates are different forms of dynamically arrested states, with gels driven by tip growth via directional interactions whereas aggregates driven by monomer addition at interior sites to maximize valency. For material properties, I highlight the crucial roles of the stress relaxation time, which is determined by the mean lifetime of intermolecular bonds in a condensate. This relaxation time dictates how the condensate manifests viscoelasticity, including shear thickening and shear thinning, and accounts for the wide variation in zero-shear viscosity among different condensates.

PhyloFrame is a Python library for phylogenetic computation targeting the gap between specialist, compiler-optimized operations and flexible, script-based workflows -- with emphasis on fast, memory-efficient operations for very large tree sizes (e.g., $\geq$ 300,000 taxa). PhyloFrame is built around a DataFrame-based tree representation, where each row corresponds to a node and columns record ancestor relationships, branch lengths, taxon labels, and any user-defined attributes. Crucial for scalability, such array-backed storage allows both library and end-user code alike to seamlessly harness Just-in-Time (JIT) compilation (e.g., Numba) and vectorized execution (e.g., NumPy, Polars). At large tree sizes, performance generally matches or exceeds Python libraries backed by native code -- notably, achieving strong performance in topological-order traversals and Newick I/O. DataFrame-based representation affords several additional conveniences, including: - succinct bulk operations (e.g., NumPy); - powerful queries and transformations (e.g., Polars expressions, Pandas indexing, SQL-style joins and merges); - compatibility with modern tabular data formats that are compression-friendly, type-aware, nullable, and highly portable (e.g., Parquet); and - broad interoperation with table-oriented data science tools (e.g., Seaborn, Plotly, Vega-Altair, tidyverse, Excel). Current library features include tree input/output, synthetic tree generation, taxon-based queries, tree traversals, tree metrics, tree manipulation, tree downsampling, and tree comparison. Most functionality supports both Pandas and Polars DataFrames, and is available through programmatic and CLI-based interfaces.

Single-cell RNA sequencing (scRNA-seq) profiles large numbers of cells but loses spatial context, whereas spatial transcriptomics (ST) preserves partial spatial structure at lower resolution. Most existing integration methods either deconvolve spot mixtures or map cells onto a measured spot lattice, which ties reconstructions to a fixed grid and slide-specific coordinate systems, a limitation that is especially problematic in unpaired settings. We propose GEARS, a geometry-first framework that reconstructs an intrinsic single-cell spatial geometry guided by ST, without relying on cell-type labels, histological images, or cell-to-spot assignment. GEARS first learns a domain-invariant expression encoder that aligns ST spots and dissociated cells, and then trains a permutation-equivariant generator with a diffusion-based refiner with EDM-style preconditioning to generate local spatial geometries under pose-invariant supervision derived from ST coordinates. At inference, GEARS reconstructs geometry on many overlapping subsets of scRNA-seq cells, aggregates predicted pairwise distances across subsets, and solves a global distance-geometry problem to obtain canonical two-dimensional coordinates and a dense distance matrix. Extensive quantitative and qualitative experiments, including cross-section generalization, show that GEARS consistently improves global distance preservation, local neighborhood fidelity, and spatial distribution alignment compared to strong spatial mapping and deconvolution baselines.

Large EEG Foundation Models (FMs) have shown great potential for decoding EEG signals across diverse cognitive tasks. However, existing EEG-FM studies exhibit three critical limitations: opaque supervised baseline tuning, unverified contributions of complex learning paradigms, and a lack of transparency in model decision-making. To address these, we propose EEG-FM-Audit, a comprehensive evaluation and analysis pipeline designed to systematize the assessment of EEG-FMs. EEG-FM-Audit consists of three primary components: (1) an ASHA-driven benchmarking protocol that ensures fair comparisons by transparently optimizing supervised baselines; (2) paradigm-level ablation studies to evaluate the effectiveness of learning paradigms in FMs; and (3) a neurophysiological probing (NPP) framework, which explores whether FMs leverage valid temporal, spatial, and spectral EEG properties. We apply EEG-FM-Audit to four state-of-the-art EEG-FMs and five representative supervised models across three public datasets. Our results reveal that properly tuned supervised baselines can match or outperform advanced FMs, despite requiring significantly fewer parameters. Furthermore, we find that the effectiveness of learning paradigms of FMs is highly dependent on dataset scale and architecture. Finally, NPP analysis demonstrates how FMs rely on specific physiological features, establishing a framework for more interpretable neural decoding.

Messenger RNA (mRNA) sequences as therapeutics require optimized design to ensure efficient translation, structural stability, and minimal immunogenicity. This study presents a two-stage in-silico framework that integrates deep learning and evolutionary computation for rational mRNA optimization instead of existing state-of-the-art models. In the first stage, a pretrained CodonTransformer (BERT-like Large Language Model) generates biologically coherent mRNA sequences encoding the target antigen. In the second stage, a genetic algorithm (GA) evolves these candidate sequences through codon-aware crossover and synonymous mutation guided by human codon usage preferences. Fitness functions for evaluation combined translation-related metrics (CAI, tAI, codon-pair bias), mRNA structural stability (local and global MFE via RNAfold, GC content), and reduced immunogenicity (CpG/UpA motif frequency). Over successive generations (38th, 40th, and 42nd), the GA improved (achieved CAI values of 0.73 to 0.74 and tAI values of 0.63 to 0.64) CAI and tAI by over 6% and codon-pair bias is high and consistent (0.97 ) and improved ribosomal accessibility at the 5' end, with an unpaired_30 fraction reaching 0.87; Global Minimum Free Energy (MFE) converged to a balanced range of -346 to -356 kcal/mol, achieving approximately 84% base-paired structural stability, and reduced immune-stimulatory motifs - lowering the average immune penalty to 27.3 in the final generation. Linear Design produces hyper-stable transcripts (MFE < - 2000 kcal/mol) that risk translation inefficiency due to extreme rigidity, and BiLSTM-CRF focuses solely on high CAI (0.96 to 0.98) without structural constraints, our framework achieves an optimal translation-stability equilibrium, highlighting the proposed BERT-GA framework as an effective, data-driven approach for the design and optimization of in-silico mRNA sequences.

Active sensing links behavior and learning through an action-perception loop: actions determine the observations used to update internal predictive models of perception, which subsequently guide the next actions. Predictive-coding frameworks provide a natural way to model this process, since internal representations are continuously updated to predict future observations. Here, we ask how exploratory and exploitative behavioral strategies shape these internal predictive representations. We build an online learning agent in a tree-like maze with a controllable parameter regulating the balance between exploratory and exploitative regimes. The agent updates a predictive-coding-based perception model from experience generated by its own behavior. The model predicts both future maze states and reward probability, allowing the agent to select actions either by expected information gain during exploration or by predicted reward during exploitation. We show that the resulting internal predictive representations depend strongly on the agent's behavioral regime. Exploratory agents develop representations that are more spatially organized and better preserve the structure of maze transitions in latent space. In contrast, exploitative agents learn less organized representations. We then train this predictive model on natural trajectories of water-deprived mice navigating the same maze and compare the resulting representations with those learned from agent trajectories. More exploratory mice show representational geometries that closely match those of exploratory agents, whereas mice with more restricted visitation patterns resemble reward-driven, exploitative agents. Together, these findings suggest that exploration enables predictive models to form generalized internal representations by organizing latent space around both spatial location and transition context in artificial agents and animals.

Predicting whether two drugs interact (binary detection) is a substantially dif- ferent task from predicting the mechanism type of that interaction (multi-class classification). This study presents a systematic ablation study of three Graph Neural Network (GNN) architectures for drug-drug interaction (DDI) prediction on a publicly available benchmark dataset comprising 38,337 positive pairs across 86 interaction types. Three architectures are compared under identical training conditions (n = 61,339 pairs): a siamese dual Message Passing Neural Network (MPNN) with concatenation (Concat), a dual MPNN with four-head cross-attention (CrossAtt), and a ternary MPNN incorporating an interaction graph (Ternary). CrossAtt improves multi-class F1-macro by +0.186 absolute (+45%) over Concat, while improving binary AUC by only +0.012 (+1.3%) - confirming that atom-level inter-molecular communication specifically enables mechanism-type classification. The ternary architecture underperforms despite equivalent training data, with its failure consistent with a training instability hypothesis. Validation on ten acetylsali- cylic acid (ASA) drug pairs, held out prior to training, demonstrates 10/10 correct DDI-type predictions for CrossAtt versus 0/10 for Ternary. Two consistent failure cases are identified across all architectures, linking to structural limits established in a companion toxicity study.

Convergent evolution provides a useful framework for testing whether independent origins of similar traits share common genetic mechanisms. Evolutionary Sparse Learning with Paired Species Contrast (ESL-PSC) is an approach to identify genes and sites associated with convergent traits from aligned sequences by fitting sparse predictive models to phylogenetically informed species contrasts. However, practical use of ESL-PSC currently requires substantial command-line fluency for data assembly, species-pair design, execution, and output interpretation. Here we present an integrated ESL-PSC analysis environment (ESL-PSC Toolkit) centered on a graphical user interface (GUI). ESL-PSC Toolkit is designed to assist users from experimental design through data interpretation without requiring extensive technical expertise. It supports guided input validation, interactive tree-based pair selection, command preview, live execution, post-run exploration of ranked genes and aligned sites, a complementary substitution-counting method, and analysis of continuous quantitative convergent traits. The computational backend has been reimplemented in Rust with many performance optimizations and parallelism, greatly reducing runtime for most analyses and enabling cross-platform packaged distributions. Downloadable GUI and CLI toolkit software packages for Mac, Windows, and Linux are available at https://github.com/John-Allard/ESL-PSC/releases/latest.

Whisking is a rhythmic and adaptive behavior that rodents use to probe and interact with their environment, and the frequency of movement reflects both sensorimotor processing and internal brain states. A robust and traditional method of whisker frequency estimation uses power spectral analysis of whisker position spanning several cycles. To improve the temporal resolution of whisker movement, we here estimate the period for each cycle, hence indirectly extracting an instantaneous frequency. We do this using markerless estimation of whisker position and identifying the peak and trough for each cycle. The cycle period is extracted, and artifacts are rejected with a ripple exclusion validator based on peak prominence and sequential amplitude filtering. The method is compared with power spectral estimation, using the Fourier transform of a temporal window of 0.5 seconds. We find that frequency estimation using a fixed window does not capture transient variability, while the cycle by cycle method recovers higher, time-resolved frequencies. The cycle by cycle approach also reveals the expected cycle-level variability. Artifact rejection through subsequence filtering removed spurious frequencies above 30 Hz, aligning refined frequencies with established physiological bounds (4 to 28 Hz). This pipeline provides an alternative solution for real time compatible frequency estimation, which better captures temporal variation at the expense of precision in frequency estimation.

Simulation of post-prandial pharmacokinetics, such as muscle protein synthesis (MPS) through mTORC1 and insulin-induced glucose uptake, is often challenging due to the computational intensity of the multi-compartmental approach. In this study, I introduce an in silico metabolic simulator that uses bi-compartmental Bateman kinetic processes, gamma-variate distributions, and finite state machine reasoning to solve temporal differential equations instantaneously, generating metabolic curves and predictions depending on input meals. The novel underlying algorithm was custom-built entirely independent of third-party libraries or external services. This original computational engine, bridging the gap between academia and the digital health sector, is integrated within a web dashboard and provided as a service via REST APIs. The average response time is approximately 135 ms with a maximum below 750 ms. The multi-dimensional model was calibrated using a Landmark Validation approach across diverse dietary conditions (Whey Protein, mixed meal, OGTT) and optimized via Grid Search. Ultimately, the system achieved a global physiologically optimal Mean Absolute Percentage Error (MAPE) of $\sim18\%$ while maintaining an algorithmic complexity of $O(n \log n)$.

Proteins perform their biological functions through three-dimensional structures encoded by amino acid sequences, and ligand-binding protein co-design requires models that generate sequence-structure compatible proteins under explicit ligand constraints. Although continuous diffusion and flow-based models support ligand-aware design in coordinate or latent spaces, existing discrete diffusion protein language models mainly operate over sequence or structure tokens without direct small-molecule conditioning. We introduce \textbf{ProtLiD$^2$}, a \textbf{Prot}ein \textbf{L}igand-conditioned \textbf{D}iscrete \textbf{D}iffusion model for protein sequence-structure co-design. ProtLiD$^2$ jointly generates amino-acid sequence and discrete structure tokens while incorporating ligand chemical and geometric information through geometry-aware cross-attention. Trained on over one million ligand-protein complexes, ProtLiD$^2$ extends masked discrete diffusion to ligand-aware functional protein design. We further propose maximum confidence-margin guided ReMask decoding, an inference-time self-correction strategy that retains confident predictions and remasks uncertain tokens. ProtLiD$^2$ improves global fold confidence over Complexa in whole-protein design, increasing TM-score from 0.672 to 0.802 and pLDDT from 64.55 to 73.00. In pocket co-design, ProtLiD$^2$ reduces active-site BB-RMSD from 3.46/3.40Å for FAIR/PocketGen to 1.97Å, and improves ligand-aware pass rates over PocketGen from 14.86% to 59.73% and from 6.08% to 23.49% under stricter docking thresholds. These results support ligand-conditioned discrete diffusion as an effective token-space framework for functional protein co-design. Code will be available at https://github.com/auroua/ProtLiD.

Digital twins of sensory cortex serve as powerful response oracles. Although prediction accuracy is the central metric by which these models are evaluated, it provides limited insight into the latent representations that support those predictions. This becomes increasingly important as digital twins are used as in silico experimental systems for stimulus design and hypothesis generation: models with similar prediction accuracy may rely on different latent representations. We address this gap by systematically probing a family of digital twins of mouse V1 trained to predict neural activity from naturalistic videos recorded in freely moving mice. The models share the same training data and neural-prediction objective, but differ in visual-encoder architecture. For each frozen model, we characterize latent representations along three levels: (i) linear decodability from controlled visual probes of orientation, contrast, and motion; (ii) latent-unit tuning to canonical visual features including orientation selectivity, contrast response, spatial-frequency tuning; and (iii) population geometry of hidden-layer activity. Across architectures, better neural-response prediction correlates with stronger probe accuracy. Additionally, highly predictive models exhibit flatter hidden-population eigenspectra, indicating higher-dimensional representations closer to population-geometry signatures reported in mouse V1. Although these representational properties covary with prediction accuracy across architectures, digital twins with comparable prediction scores can still differ substantially in probe performance and latent-unit tuning. These results establish multi-level representational probing as a complement to standard neural-prediction evaluation, providing a framework for understanding digital twins not only as predictors, but also as substrates for studying visual computations.

Liquid chromatography-mass spectrometry (LC-MS/MS) data analysis requires adaptable software solutions to meet diverse analytical needs. We present eMZed 3, a modern Python framework for flexible and interactive analysis of LC-MS/MS data. eMZed 3 enables users to develop scalable workflows tailored to their specific requirements while leveraging Python's extensive ecosystem of libraries. Building on its predecessor, eMZed 3 is now Python 3-based and includes substantial enhancements, including support for chromatogram-based LC-MS data, a new SQLite-based backend supporting optional out-of-memory processing, and rich interactive visualization tools. Compared to the previous version, eMZed 3 is now split into three packages: emzed (core functionalities), emzed-gui (interactive data visualization), and emzed-spyder (an integrated development environment). This modular architecture allows straightforward integration of the emzed core library into headless Python environments, including computational notebooks (such as Jupyter) or high-performance computing clusters. eMZed 3 incorporates well-established libraries such as OpenMS, and is highly suited for both targeted and untargeted metabolomics. Overall, eMZed 3 supports the efficient development of scalable and reproducible LC-MS data analysis and is accessible to both novice and advanced programmers. Availability and Implementation: eMZed 3 and its documentation are freely available at https://emzed.ethz.ch, the source code is hosted at https://gitlab.com/groups/emzed3. An online-executable example workflow is available on Binder at: https://mybinder.org/v2/gl/emzed3%2Femzed-example-workflow/HEAD?labpath=example.ipynb.

Speech neuroprosthesis systems decode intended speech from neural activity in the absence of audible output, offering a path to restoring communication for individuals with speech-impairing conditions. Current approaches decode predominantly from motor cortical areas, discarding others -- such as area 44, part of Broca's area -- that may encode complementary linguistic information. We introduce MoDAl (Modality Decorrelation and Alignment), a framework that discovers complementary neural modalities through the interplay of two objectives in a shared projection space. A contrastive loss aligns each of several parallel brain encoders with the text embeddings of a pretrained large language model (LLM), while a decorrelation loss prevents the encoders from coalescing to duplicative representations. We prove that these objectives are in productive tension: Contrastive alignment induces transitive modality coalescence, which decorrelation must counteract for the framework to discover diverse neurolinguistic modalities. On the Brain-to-Text Benchmark '24, MoDAl reduces word error rate (WER) from 26.3% to 21.6% compared to the previous best end-to-end method, with the gain from incorporating previously discarded area 44 signals arising entirely from the decorrelation mechanism. Analysis of the discovered modalities reveals functional specialization: Encoders receiving area 44 input capture structural and syntactic properties (sentence length, grammatical voice, wh-words), consistent with the neurolinguistic understanding of Broca's area.

Ultraweak photon emission, also referred to as biological autoluminescence or biophoton emission, is the spontaneous emission of extremely low levels of light from a broad range of biological systems. Recent studies have reported that UPE measured extracranially can serve as a potential non-invasive biomarker of brain activity. Here, we show that this interpretation suffers from serious problems. We show that, when observed under properly dark conditions, the UPE from the head is much weaker than what is reported in certain papers on 'brain UPE' from human heads. We also show that the large signals reported in these studies can be explained by background light contamination. Furthermore, photons with wavelengths < 600 nm are strongly attenuated by scalp and skull tissues, and longer wavelengths fall largely outside the effective spectral sensitivity of the photomultiplier tubes (PMTs) used. As a consequence, even if UPE from the head is detected under properly background-free conditions, it is likely to be dominated by emission from the scalp rather than from the brain, certainly as long as PMTs are used. Our results emphasize the importance of careful experimental design to make genuine progress on this important question.

Species sharing a habitat will co-evolve to make use of the available resources, as consumption is modulated by competition and negative feedback loops between consumers and resources. The dietary range of a given species determines the resources it has access to and thus the other species with which it competes. A narrow dietary range avoids competition at the cost of over-reliance on a small selection of resources; conversely a wide dietary range provides more alternatives but also more chance of competition with other species. Here, we investigate the evolution of dietary range within a mathematical model of niche formation. We find highly path dependent co-evolution dynamics characterised by long-lived quasi-stable states. Ultimately, stochastic effects drive the evolution of generalist diets, as we uncover in our analysis and simulations.

Common deep learning approaches for antibody engineering focus on modeling the marginal distribution of sequences. By treating sequences as independent samples, however, these methods overlook affinity maturation as a rich and largely untapped source of information about the evolutionary process by which antibodies explore the underlying fitness landscape. In contrast, classical phylogenetic models explicitly represent evolutionary dynamics but lack the expressivity to capture complex epistatic interactions. We bridge this gap with CoSiNE, a continuous-time Markov chain parameterized by a deep neural network. Mathematically, we prove that CoSiNE provides a first-order approximation to the intractable sequential point mutation process, capturing epistatic effects with an error bound that is quadratic in branch length. Empirically, CoSiNE outperforms state-of-the-art language models in zero-shot variant effect prediction by explicitly disentangling selection from context-dependent somatic hypermutation. Finally, we introduce Guided Gillespie, a classifier-guided sampling scheme that steers CoSiNE at inference time, enabling efficient optimization of antibody binding affinity toward specific antigens.

Lineage marker population frequencies can serve as one way to express evidential value in forensic genetics. However, for high-quality whole mitochondrial DNA genome sequences (mitogenomes), population data remain limited. In this paper, we offer a new method, MitoFREQ, for estimating the population frequencies of mitogenomes. MitoFREQ uses the mitogenome resources HelixMTdb and gnomAD, harbouring information from 195,983 and 56,406 mitogenomes, respectively. Neither HelixMTdb nor gnomAD can be queried directly for individual mitogenome frequencies, but offers single nucleotide variant (SNV) allele frequencies for each of 30 "top-level" haplogroups (TLHG). We propose using the HelixMTdb and gnomAD resources by classifying a given mitogenome within the TLHG scheme and subsequently using the frequency of its rarest SNV within that TLHG weighted by the TLHG frequency. We show that this method is guaranteed to provide a higher population frequency estimate than if a refined haplogroup and its SNV frequencies were used. Further, we show that top-level haplogrouping can be achieved by using only 227 specific positions for 99.9% of the tested mitogenomes, potentially making the method available for low-quality samples. The method was tested on two types of datasets: high-quality forensic reference datasets and a diverse collection of scrutinised mitogenomes from GenBank. This dual evaluation demonstrated that the approach is robust across both curated forensic data and broader population-level sequences. This method produced likelihood ratios in the range of 100-100,000, demonstrating its potential to strengthen the statistical evaluation of forensic mtDNA evidence. We have developed an open-source R package `mitofreq` that implements our method, including a Shiny app where custom TLHG frequencies can be supplied.

Dynamical systems models such as recurrent neural networks (RNNs) are increasingly popular in theoretical neuroscience for hypothesis-generation and data analysis. Evaluating the dynamics in such models is key to understanding their learned generative mechanisms. However, such evaluation is impeded by two major challenges: First, comparison of learned dynamics across models is difficult because there is no enforced equivalence of their coordinate systems. Second, identification of mechanistically important low-dimensional motifs (e.g., limit sets) is intractable in high-dimensional nonlinear models such as RNNs. Here, we propose a comprehensive framework to address these two issues, termed Diffeomorphic vector field alignment FOR learned Models (DFORM). DFORM learns a nonlinear coordinate transformation between the state spaces of two dynamical systems, which aligns their trajectories in a maximally one-to-one manner. In so doing, DFORM enables an assessment of whether two models exhibit topological equivalence, i.e., similar mechanisms despite differences in coordinate systems. A byproduct of this method is a means to locate dynamical motifs on low-dimensional manifolds embedded within higher-dimensional systems. We verified DFORM's ability to identify linear and nonlinear coordinate transformations using canonical topologically equivalent systems, RNNs, and systems related by nonlinear flows. DFORM was also shown to provide a quantification of similarity between topologically distinct systems. We then demonstrated that DFORM can locate important dynamical motifs including invariant manifolds and saddle limit sets within high-dimensional models. Finally, using a set of RNN models trained on human functional MRI (fMRI) recordings, we illustrated that DFORM can identify limit cycles from high-dimensional data-driven models, which agreed well with prior numerical analysis.

Predicting future brain state from a baseline magnetic resonance image (MRI) is a central challenge in neuroimaging and has important implications for studying neurodegenerative diseases such as Alzheimer's disease (AD). Most existing approaches predict future cognitive scores or clinical outcomes, such as conversion from mild cognitive impairment to dementia. Instead, here we investigate longitudinal MRI image-to-image prediction that forecasts a participant's entire brain MRI several years into the future, intrinsically modeling complex, spatially distributed neurodegenerative patterns. We implement and evaluate five deep learning architectures (UNet, U2-Net, UNETR, Time-Embedding UNet, and ODE-UNet) on two longitudinal cohorts (ADNI and AIBL). Predicted follow-up MRIs are directly compared with the actual follow-up scans using metrics that capture global similarity and local differences. The best performing models achieve high-fidelity predictions, and all models generalize well to an independent external dataset, demonstrating robust cross-cohort performance. Our results indicate that deep learning can reliably predict participant-specific brain MRI at the voxel level, offering new opportunities for individualized prognosis.

Motivation: Advances in high-throughput chromatin conformation capture have provided insight into the three-dimensional structure and organization of chromatin. While bulk Hi-C experiments capture spatio-temporally averaged chromatin interactions across millions of cells, single-cell Hi-C experiments report on the chromatin interactions of individual cells. Supervised and unsupervised algorithms have been developed to embed single-cell Hi-C maps and identify different cell types. However, single-cell Hi-C maps are often difficult to cluster due to their high sparsity, with state-of-the-art algorithms achieving a maximum Adjusted Rand Index (ARI) of only < 0.4 on several datasets while requiring labels for training. Results: We introduce a novel unsupervised algorithm, Single-cell Clustering Using Diagonal Diffusion Operators (SCUDDO), to embed and cluster single-cell Hi-C maps. We evaluate SCUDDO on three previously difficult-to-cluster single-cell Hi-C datasets, and show that it can outperform other current algorithms in ARI by > 0.2. Further, SCUDDO outperforms all other tested algorithms even when we restrict the number of intrachromosomal maps for each cell type and when we use only a small fraction of contacts in each Hi-C map. Thus, SCUDDO can capture the underlying latent features of single-cell Hi-C maps and provide accurate labeling of cell types even when cell types are not known a priori. Availability: SCUDDO is freely available at www.github.com/lmaisuradze/scuddo. The tested datasets are publicly available and can be downloaded from the Gene Expression Omnibus.

Recently, many generative models for de novo protein structure design have emerged. Yet, only few tackle the difficult task of directly generating fully atomistic structures jointly with the underlying amino acid sequence. This is challenging, for instance, because the model must reason over side chains that change in length during generation. We introduce La-Proteina for atomistic protein design based on a novel partially latent protein representation: coarse backbone structure is modeled explicitly, while sequence and atomistic details are captured via per-residue latent variables of fixed dimensionality, thereby effectively side-stepping challenges of explicit side-chain representations. Flow matching in this partially latent space then models the joint distribution over sequences and full-atom structures. La-Proteina achieves state-of-the-art performance on multiple generation benchmarks, including all-atom co-designability, diversity, and structural validity, as confirmed through detailed structural analyses and evaluations. Notably, La-Proteina also surpasses previous models in atomistic motif scaffolding performance, unlocking critical atomistic structure-conditioned protein design tasks. Moreover, La-Proteina is able to generate co-designable proteins of up to 800 residues, a regime where most baselines collapse and fail to produce valid samples, demonstrating La-Proteina's scalability and robustness.

Multicellular organisms contain a wide variety of highly specialized cell types. The consistency and robustness of developmental trajectories suggest that complex gene regulatory networks effectively act as low-dimensional cell fate landscapes. Prior work inspired by dynamical systems theory argues that cell fate transitions fall into universal decision-making classes, but the theory connecting these geometric landscapes to high-dimensional gene expression space is still in its infancy. Here, we introduce a phenomenological model that identifies experimental signatures of decision-making classes in single-cell RNA-sequencing time-series data. The model combines low-dimensional gradient-like dynamics with high-dimensional Hopfield networks to capture the interplay between cell fate, gene expression, and signaling. We apply the framework to experimental mouse data on maturing lung alveolar cells and lineage-traced hematopoietic differentiation and show that the measured cell fate dynamics are consistent with developmental landscapes containing intermediate progenitors and saddle points. We further show that the framework can be used to understand spatial patterning and cell fate organization, focusing on Notch signaling in lung airways. Together, these results provide evidence that collective transcriptomic dynamics carry signatures of landscape features associated with universal decision-making classes.

Proteins play a vital role in biological processes and are indispensable for living organisms. Accurate representation of proteins is crucial, especially in drug development. Recently, there has been a notable increase in interest in utilizing machine learning and deep learning techniques for unsupervised learning of protein representations. However, these approaches often focus solely on the amino acid sequence of proteins and lack factual knowledge about proteins and their interactions, thus limiting their performance. In this study, we present GOProteinGNN, a novel architecture that enhances protein language models by integrating protein knowledge graph information during the creation of amino acid level representations. Our approach allows for the integration of information at both the individual amino acid level and the entire protein level, enabling a comprehensive and effective learning process through graph-based learning. By doing so, we can capture complex relationships and dependencies between proteins and their functional annotations, resulting in more robust and contextually enriched protein representations. Unlike previous methods, GOProteinGNN uniquely learns the entire protein knowledge graph during training, which allows it to capture broader relational nuances and dependencies beyond mere triplets as done in previous work. We perform a comprehensive evaluation on several downstream tasks demonstrating that GOProteinGNN consistently outperforms previous methods, showcasing its effectiveness and establishing it as a state-of-the-art solution for protein representation learning.