Latest bioRxiv papers

Category: bioinformatics — Showing 50 items

• HERMES: Holographic Equivariant neuRal network model for Mutational Effect and Stability prediction

  Visani, G. M.; Jones, Z.; Galvin, W.; Pun, M. N.; Daniel, E.; Borisiak, K.; Wagura, U.; Nourmohammad, A.

  Abstract

  Accurately predicting how amino acid substitutions alter protein function is a central challenge in biology, with applications from interpreting disease variants to designing vaccines and therapeutics. We introduce HERMES, a family of fast, structure-based models that predict mutational effects from the local atomic environment around each residue. Pre-trained on masked amino acid prediction, HERMES shows strong zero-shot performance for predicting changes in thermodynamic stability and protein-protein binding affinity. Analyzing its predictions, we uncover a pre-training bias toward size-conserving substitutions, which we reduce through an amortized fine-tuning strategy that incorporates packing flexibility. When fine-tuned on experimental data, HERMES matches state-of-the-art stability predictors without costly data augmentation. HERMES also identifies antigen-stabilizing mutations across multiple viral envelope proteins, enabling an efficient pipeline for designing mutation libraries for vaccine development. Together, this work establishes HERMES as a fast and practical structure-based framework for mutation screening and offers insight into the mechanisms underlying its predictions.

  bioinformatics2026-08-05v5

• FlashS reveals multiscale spatial gene programs at atlas scale

  Yang, C.; Zhang, X.; Chen, J.

  Abstract

  Spatial transcriptomics links gene expression to tissue architecture, but detecting spatially variable genes at atlas scale remains difficult because biologically relevant patterns are multiscale, sparse, and often non-parametric. Here we show that FlashS, a frequency-domain kernel test using random Fourier features, sparse sketching, and a kurtosis-corrected null, retains Gaussian-kernel flexibility at near-linear per-gene cost without permutation. Across 50 benchmark datasets spanning 9 spatial transcriptomics platforms, FlashS achieves the highest ranking accuracy among 14 compared methods while maintaining calibrated inference under simulation and full-atlas permutation. In human heart tissue, it recovers a mitochondrial biogenesis program that co-localizes with ventricular cardiomyocytes. On the Allen Brain MERFISH atlas containing 3.94 million cells, FlashS completes in minutes on a single workstation and separates biological signal from negative-control barcodes even when p-values saturate.

  bioinformatics2026-08-05v4

• IMAS resolves perturbation-sensitive regulatory architectures from matched tumour multiomics

  Deyang, W.; Yamashiro, T.; Inubushi, T.

  Abstract

  Tumour single-cell datasets contain weak, sparse and context-restricted regulatory signals that are difficult to distinguish from noise using expression measurements alone. Here we present IMAS, an integrative multiomic augmentation system that learns transferable regulatory structure from a pan-cancer foundation of matched single-cell RNA and chromatin-accessibility profiles and adapts it to data-limited target datasets. We refer to the resulting target-specific regulatory architectures as multi-layer target dependencies (MLTDs). MLTDs prioritize signals that remain supported across coordinated molecular, communication and perturbation-sensitive evidence, rather than by expression magnitude or any single prediction score. Rather than replacing the observed expression matrix, IMAS adds an interpretable regulatory-support layer for mechanism discovery. Across independent tumour datasets, IMAS preserved matched cross-layer supervision, concentrated predictive support into compact target-aligned structures and improved recovery of RNA and transcription-factor states. In colorectal cancer, MLTDs resolved a SOD2-associated perturbation-sensitive architecture that was distinct from native expression, conventional co-expression and the inherited pan-cancer hierarchy. These dependencies promoted propagation of perturbation-associated information through RNA-TF-regulatory-element bridges and into receiver-TF-aware communication across malignant cell states. A LAMB1-centred analysis further showed that successive regulatory, communication and temporal constraints restored an expected extracellular-matrix programme that was weakly represented in the original matrix. In head and neck squamous cell carcinoma, SOX2-centred MLTDs resolved malignant-state-specific perturbation programmes. In renal cancer, MLTD-guided analysis identified tumour-vascular coupling associated with spatially localized endothelial-to-mesenchymal-transition-like states in Xenium data. Together, IMAS reframes tumour multiomic augmentation as the recovery of compact, target-specific regulatory architectures rather than expression-matrix completion, providing an interpretable framework for prioritizing experimentally tractable mechanisms in heterogeneous tumour systems.

  bioinformatics2026-08-05v4

• Cross-Platform Calibration of Epigenetic Age Between the EPICv2 and MSA Arrays

  Tomo, Y.; Shoji, T.; Nakaki, R.

  Abstract

  Epigenetic clocks are regression models that predict biological age based on the DNA methylation patterns. They are developed using DNA methylation array data; however, systematic differences in the predicted values between the Infinium Methylation Screening Array (MSA) and Infinium MethylationEPIC v2.0 BeadChip (EPICv2) platforms have not been evaluated. We quantified the systematic differences between the MSA and EPICv2 arrays for six major epigenetic clocks (Horvath, Hannum, PhenoAge, GrimAge, GrimAge v2, and DunedinPACE) using 166 human blood samples measured on both platforms. For the five clocks expressed in years, the mean differences (MSA - EPICv2) ranged from -7.67 years for the Horvath clock to -0.634 years for GrimAge v2, despite strong correlations between platforms (r = 0.946 -- 0.991). For DunedinPACE, which predicts the pace of aging, the mean difference was -0.117, with a correlation coefficient of 0.855 between the platforms. Based on these training data, we developed three linear regression models to map the MSA-based epigenetic ages and pace of aging to the EPICv2-based values, which were used as a reference: Model 1 (offset location calibration), Model 2 (slope and location calibration), and Model 3 (slope and location calibration with covariates). Validation using 48 independent samples measured at separate institutions showed that the calibration models reduced cross-platform differences for Horvath, Hannum, PhenoAge, GrimAge, and DunedinPACE. Although Model 3 yielded mean differences closest to zero for the Horvath clock (-0.227 years) and PhenoAge (-0.510 years), Model 1, the simplest specification, also reduced the mean cross-platform differences to -1.03 years for the Horvath clock and -1.37 years for PhenoAge. For DunedinPACE, Model 1 reduced the mean difference from -0.092 to 0.025 and the RMSE from 0.103 to 0.052. These methods are expected to enable reliable epigenetic age comparisons across platforms and facilitate large-scale epigenetic studies.

  bioinformatics2026-08-05v2

• LOCALE: Local-Alignment Embeddings for Noise-Robust DNA Search at SRA Scale

  Synk, R.; Pandey, P.; Sahinalp, C.; Duraiswami, R.

  Abstract

  Searching petabase-scale repositories of raw sequencing data such as the NIH Sequence Read Archive (SRA) could transform biological discovery, but existing methods either do not scale well or rely on exact k-mer matching that is brittle to sequencing errors and biological divergence. We recast sequence search as dense retrieval: we learn vector embeddings whose inner-product similarity ranks locally aligned sequences above unaligned ones. Our key observation is that effective retrieval does not require accurate regression of global edit distance---it only requires that sequences with better local alignments score higher than sequences with worse ones. We train a DNABERT-2 encoder with an InfoNCE objective on biologically informed augmentations: overlapping crops of parent sequences corrupted with substitutions, insertions, and deletions. On a 50-accession SRA benchmark, LOCALE maintains 62.4% average Recall@Rq at a 10% mutation rate, while every baseline we evaluated falls below 60% Recall@Rq in the noisy-query setting. The advantage holds at scale: on a 500-accession, 15-Gbp benchmark, LOCALE achieves AUPRC 0.508 at 10% mutation versus 0.129 for MetaGraph.

  bioinformatics2026-08-05v2

• Scientific computing in the age of agentic AI: an exploratory field report

  Li, J.; Rubinsteyn, A.; Feldman, S.; O'Donnell, T.; Ferguson, J. M.; Patro, R.; Driver, I.; Ewels, P. A.; Krueger, F.; Angerer, P.; Gold, I.; Manning, J.; Heumos, L.; Ahangari, M.; Goyal, V.; Masoudi, H.; Pedersen, B.; Bai, A.; Li, H.; Shringarpure, S.; Myers, E. W.; Ho, A.

  Abstract

  Scientific computing has become a central component of modern scientific discovery. Yet many computational tools are developed by small, specialized teams under incentives that encourage the release of rapidly prototyped tooling without commensurate attention to engineering concerns, including performance and maintainability. These gaps are particularly visible in the life sciences, where the advent of high-throughput sequencing and molecular profiling has made the production and processing of datasets routine at scales that strain reliability and cost. Recently, LLM-based agents have become increasingly capable, with publicly available systems possessing both significant domain knowledge in many scientific fields and the ability to autonomously operate over complex and specialized codebases in pursuit of well-defined goals. Together, these developments create a practical opportunity for scientific computing. Many of the persistent weaknesses of the scientific computing ecosystem stem from technical debt and a shortage of sustained engineering labor and expertise. Here, we examine coding agents as a potential way to address these weaknesses: we present an exploratory field report of eight early case studies in the application of LLM agents to scientific computing across a range of project scopes, from lightweight maintenance tasks to full performance-oriented rewrites of scientific libraries, with a focus on the life sciences. Each of these case studies is accompanied by reflections from the individual or group responsible for the work, including lessons from the process. Overall, we find that the use of coding agents in scientific computing holds great promise for accelerating scientific research and increasing the reliability of critical systems, but that outstanding concerns remain, including responsibility and ownership for such projects, and we suggest collaboration and stewardship with existing maintainers when feasible.

  bioinformatics2026-08-05v2

• MolX: A Geometric Foundation Model for Protein-Ligand Modelling

  Liu, J.; Pan, T.; Guo, X.; Ran, Z.; Hao, Y.; Yang, Y.; Ng, A. P.; Pan, S.; Song, J.; Li, F.

  Abstract

  Understanding how small molecules interact with protein binding pockets is central to structure-based drug discovery. Accurately modelling these interactions requires capturing the 3D geometry and physicochemical complementarity of binding interfaces, yet existing computational approaches encode proteins and ligands separately or rely on simplified structural representations that do not explicitly model cross-entity spatial relationships. Such decoupled representations restrict their capacity to capture interface-level geometric constraints that arise from protein-ligand co-organisation. Here we present MolX, a Graph Transformer foundation model that jointly learns geometric and chemical representations of protein pockets and ligands from large-scale 3D structural data. Integrating over 3 million protein pockets and 5 million molecules, MolX represents both entities as E(3)-equivariant graphs to preserve spatial geometry and chemical context. The architecture employs dual E(3)-equivariant graph Transformer encoders to model pocket and ligand embeddings, ensuring representations remain invariant to rotation, translation, and reflection. MolX is pretrained using a hybrid learning paradigm that combines supervised biochemical objectives, logP and energy-gap regression, with self-supervised geometric objectives, coordinate reconstruction, and atom-type prediction, fostering generalisable molecular understanding. Across eight downstream benchmarks, including antibody-drug conjugates (ADC), proteolysis-targeting chimeras (PROTAC), molecular glue, and PCBA activity prediction, as well as binding affinity and physicochemical property regression, MolX achieves consistent state-of-the-art performance and strong cross-domain generalisation. Furthermore, MolX incorporates a sparse autoencoder module to decompose latent representations into interpretable biological components, thereby revealing the pocket-ligand interactions that drive prediction outcomes. Together, MolX establishes a scalable and interpretable foundation model for molecular representation learning, providing a unified framework for predicting and interpreting complex small-molecule-protein interactions.

  bioinformatics2026-08-05v2

• Graph Machine Learning for Physiological Role Prediction in Protein Contact Networks: A Large-Scale Comparative Study on the Human Proteome

  Cervellini, M.; Martino, A.

  Abstract

  Proteins are fundamental macromolecules involved in virtually all biological processes. Their physiological roles are tightly linked to their three-dimensional structure, which can be naturally abstracted as Protein Contact Networks (PCNs), i.e., graphs where residues are nodes and edges encode spatial proximity. This representation enables the application of graph machine learning to address the protein functional annotation gap at proteome scale. In this work, protein function prediction is studied on the majority of the human proteome, focusing on enzymatic activity and enzyme class assignment as well-defined and biologically meaningful targets. A large-scale supervised analysis was conducted on PCNs derived from experimentally resolved human protein structures. Multiple graph-based learning paradigms were systematically compared under a unified evaluation protocol, including handcrafted graph embeddings, kernel methods, and end-to-end Graph Neural Networks (GNNs). Feature engineering approaches included (i) spectral density embeddings of the normalized graph Laplacian and (ii) higher-order topological representations based on simplicial complexes, with optional INDVAL-based feature selection. These representations were paired with linear, ensemble, and kernel classifiers, while GNNs were trained directly on raw PCNs exploiting a diverse set of message-passing strategies. Two tasks were considered: binary classification of enzymatic versus non-enzymatic proteins and multiclass prediction of first-level Enzyme Commission (EC) classes. Performance was assessed using repeated stratified splits to ensure robust and variance-aware evaluation. In the binary enzymatic classification task, the Jaccard-based graph kernel achieved the best performance with an adjusted balanced accuracy of 0.90, closely followed by GNNs trained end-to-end on PCNs. In the multiclass EC prediction task, GNNs demonstrated superior discriminative power, reaching an adjusted balanced accuracy of 0.92 and outperforming all explicit embedding and kernel-based approaches. Overall, results indicate that EC class prediction is intrinsically more complex than binary enzymatic discrimination and benefits from the higher expressivity of deep message-passing architectures. The findings demonstrate that graph-based representations of protein structure support competitive functional prediction at proteome scale, with classical kernel methods and modern GNNs offering complementary strengths in terms of accuracy, inference-time applicability, and flexibility.

  bioinformatics2026-08-05v2

• A calibrated novelty flag for fungal ITS metabarcoding: choosing the error rate at which sequences are declared new

  O'Brien, A.; Marin, C.; Parada, P.

  Abstract

  Environmental fungal surveys routinely recover internal transcribed spacer (ITS) sequences that cannot be assigned at fine taxonomic ranks, the so-called fungal "dark matter," and in practice such sequences are set aside by thresholding a similarity or confidence score at a conventional value. Those conventions do abstain, but the error rate a given threshold implies is neither stated nor selectable, and a threshold defined for one kind of score does not transfer to another. We present a conformal novelty flag that supplies what is missing: each query receives a p-value with a distribution-free guarantee that the rate of falsely declaring a known sequence novel is bounded by a user-chosen alpha. On a leave-one-genus-out benchmark built from UNITE the flag holds its nominal rate across two orders of magnitude in alpha, so the practitioner selects an operating point on a stated frontier rather than inheriting one: at alpha=0.05 it fires on 4.3% of known-genus queries and recovers 19.2% of genuinely novel genera, and at alpha=0.20 on 19.7% and 50.9%. It calls 4,996 of 5,000 non-fungal SSU outgroup sequences novel, anchoring the far end of a monotonic known-to-outgroup gradient. The guarantee is score-agnostic in practice and not only in principle: applied to alignment identity, a k-mer bootstrap consensus and two neural classifiers' output probabilities, across two amplicon regions and a sevenfold change in reference size, all eleven calibrations fell within 1.1 percentage points of nominal. Detection over those same eleven settings ranged from 5.2% to 53.3%, a tenfold spread under an essentially constant error rate, which makes explicit that the guarantee is on the error rate and not on power: a well-calibrated flag built on an uninformative score is valid and useless, and two of our settings are exactly that. Applied to a soil fungal dataset the flag identifies 20.7% of amplicon sequence variants as novel at a controlled 5% error rate, falling to 16.3% under an abundance filter. Half of those recur near-identically among the unnamed environmental variants of GlobalFungi while fewer than one in ten matches a named species hypothesis, a sixfold skew towards the uncatalogued against 1.9-fold for the sequences the flag passes: present in soils worldwide and named in almost none of them. The flag turns an arbitrary cutoff into a decision with a stated error rate, and in doing so converts dark matter from a residue into a set of prioritizable targets. Source code and the benchmark harness are available under the MIT license at https://github.com/ayobi/its-novelty.

  bioinformatics2026-08-05v2

• Pretrained deep-learning ITS classifiers read the flanking regions, not the ITS2 barcode, and so fail on the amplicon that environmental fungal surveys sequence

  O'Brien, A.; Marin, C.; Parada, P.

  Abstract

  Deep-learning classifiers for the fungal internal transcribed spacer (ITS) report accuracies above 90% and are increasingly proposed for environmental metabarcoding. We benchmark two pretrained models, a convolutional network and a transformer sharing a training corpus of 5.23M sequences, against two established k-mer methods on 5,222 identical queries, one per genus, evaluated on both full-length ITS and the ITS2 subregion that dominates environmental sequencing. The design favours the classifiers: queries are drawn from the same public dataset they were trained on and stratified by whether a query's genus lies in their own label space, recovered from the distributed models, while the reference the k-mer methods consult mirrors that label space and excludes the queries themselves. Even so, on full-length ITS both classifiers are outperformed by both classical methods at every rank and in both strata: SINTAX recovers the correct family for 92.0% of seen-genus and 70.3% of novel-genus queries and best-hit alignment against a 56,327-sequence reference for 92.3% and 67.2%, against 77.9% and 57.7% for the transformer and 76.5% and 54.3% for the convolutional network. A hierarchical logistic regression on k-mer counts, fitted in ten minutes to 1.07% of the MycoAI training corpus, also exceeds both and places novel genera better than either search method, and refitted on ITS2 it recovers 89.2% of seen-genus families on that amplicon against 88.6% for best-hit alignment, so neither learned classification nor the amplicon is what fails. Restricting the identical records to ITS2 costs the k-mer methods 3.5 and 3.7 percentage points of seen-genus family accuracy but costs the classifiers 49.6 and 58.0, reducing them to 28.3% and 18.5%. An ablation identifies the cause. Grafting each query's unaltered ITS2 between the flanking regions of a donor record from a different phylum returns the donor's family for 34.4% of queries against the query's own for 4.2% in the convolutional model, and 63.7% against 0.8% in the transformer, from a baseline of 0.1% where no donor sequence is present. The models therefore read taxonomy principally from the flanking regions rather than from the ITS2 barcode, which explains the collapse and predicts the same failure for any subregion amplicon. Compounding this, on ITS2 the classifiers' output probability all but ceases to separate novel from known genera (AUROC 0.541 and 0.503, the latter at chance, against 0.866 for alignment identity and 0.785 for the SINTAX bootstrap), so the failure is not detectable from the models' own output. We recommend that reported accuracies for such models specify the amplicon region of the evaluation, state the length distribution of the training corpus, and include a same-query classical baseline.

  bioinformatics2026-08-05v2

• From Abandoned Scripts to FAIR Community Pipelines: Rescuing Orphan Bioinformatics Workflows with nf-core - Lessons from Light-Sheet Fluorescence Microscopy

  Schwitalla, C.; Kuhn Cuellar, L.; Hoertenhuber, M.; Grote, N.; Woller, T.; Lamberti, I.; Pavie, B.; Kuestner, T.; Kyere, F. A.; Curtin, I.; Stein, J. L.; Nahnsen, S.

  Abstract

  Background Research software is essential for modern data analysis but is often developed and maintained by a small number of researchers. When developers leave, software may become orphaned, limiting reuse and risking the loss of valuable domain knowledge and computational methods. While the FAIR Principles for Research Software (FAIR4RS) provide an essential foundation for improving the reuse of research software, compliance with these principles alone does not guarantee practical reusability. Here, we investigate whether orphaned scientific software can be systematically rescued and transformed into sustainable, reusable workflows using established software engineering practices and community standards. Findings We re-engineered the abandoned MATLAB-based NuMorph toolkit for large-scale light-sheet microscopy image analysis into nf-core/lsmquant, a Nextflow-based workflow developed according to nf-core community guidelines. The re-engineered workflow preserved the original scientific methods at comparable computational cost while improving the software's FAIRness, portability, and reproducibility. Integration into the nf-core ecosystem provides a community-driven framework that supports software sustainability through distributed maintenance and shared development practices, while the modular workflow architecture simplified adaptation of nf-core/lsmquant to additional light-sheet microscopy datasets beyond the original application Conclusion Our work demonstrates that orphaned scientific software can be successfully rescued through systematic re-engineering guided by FAIR and software sustainability principles. By transforming a legacy codebase into a community-maintained workflow, we preserve valuable domain-specific methods while improving usability, maintainability, and reproducibility. This approach provides a practical strategy for recovering orphan research software and integrating it into modern, reusable research ecosystems. Keywords Light-sheet fluorescence microscopy, Nextflow, nf-core, re-engineering, sustainable software

  bioinformatics2026-08-05v2

• BirdCODE: Detecting bird communication at scale

  Fine, A.; Hoffman, B.; Robinson, D.; Miron, M.; Alizadeh, M.; Chemla, E.; Cusimano, M.; James, L. S.; Keen, S.; Matthies, E.; Narula, G.; Nolasco, I.; Geist, M.

  Abstract

  Deep learning-based animal sound identification is regularly applied to large audio datasets for ecological monitoring and citizen science, but existing methods lack the fine temporal resolution required to extract insights into animal communication from these same datasets. Here we introduce Bird Communication Detector (BirdCODE), a deep learning model that detects and classifies the vocalizations of over 9000 bird species with precise temporal boundaries, a several hundred-fold increase the number of species over previous bioacoustic sound event detection models. In extensive benchmarking, BirdCODE achieves state-of-the-art performance in detection and classification of bird sounds. Applying BirdCODE to 1.3M citizen-science recordings, we present four case studies of how BirdCODE-computed sound event boundaries can be used to carry out phylogenetic analyses, to describe geographic and temporal variation in acoustic communication, and to characterize cross-species interactions. Together, these demonstrate how BirdCODE can enable large-scale, data-driven studies of bird communication. Model code, weights, and predictions are publicly available.

  bioinformatics2026-08-05v1

• Contrastive Regulatory Embeddings Attention Model for Differential Expression Prediction with Personalized Genomes: Insights and Challenges

  Hu, Z.; Ku, J.; Pollard, K.

  Abstract

  Deep learning models applied to DNA sequences have achieved success in predicting gene expression, chromatin profiles, and variant pathogenicity. However, learning cross-individual differences remains challenging because DNA sequence variation between individuals is small, while gene expression is heavily influenced by non-genetic noise. Prior efforts to predict personalized gene expression from sequence have shown limited generalizability beyond training genes, revealing limitations such as the dilution of variant signals among highly similar input sequences across consecutive convolutional downsampling layers. In this work, we explore architectural modifications addressing these challenges. We propose a contrastive regulatory embedding attention model (CREAM) designed to better capture subtle sequence differences between individuals. To mitigate the impact of non-genetic variability, we decompose gene expression into genetic and non-genetic components and evaluate model performance on the genetic signal. Evaluated on simulated and GTEx transcriptomic datasets, CREAM captures tissue-specific gene expression and significantly outperforms baseline architectures on training genes. CREAM autonomously prioritizes statistically fine-mapped causal expression quantitative trait loci, and introducing an auxiliary $L_1$ inductive bias further sharpens localizing causal variant in unseen genes. However, generalization to predicting expression of unseen genes collapses to near-zero correlation among all tested methods. Single-run model predictive uncertainty capture prediction accuracy in training genes and mirrors cross-run consistency in test genes. While accurate inference on unseen genes remains an open problem, our results highlight key obstacles and suggest directions for modeling personalized gene regulation from DNA sequence.

  bioinformatics2026-08-05v1

• Multi-modal Graph Integration for Biologically Interpretable Domain Identification Using Spatial Transcriptomics

  Lee, S.; Wang, G.; Kang, K.; Chen, J.; Kang, M.

  Abstract

  Functional domain identification in spatial transciptomics transforms spatial molecular measurements into mechanistic insights into tissue physiology and pathology. However, the inherent noise and sparsity of gene expression data, along with the locality-biased design of conventional graph-based approaches, fundamentally limit the accurate identification of complex tissue domains. In this study, we propose a novel Biologically Interpretable multi-modal Graph using Spatial Transcriptomics, called BIGraph-ST, that integrates pathway activity scores and histological image features for robust spatial domain identification. BIGraph-ST represents modality-specific similarity through affinity graphs and propagates spatial topology to capture higher-order connectivity within the tissue microenvironment. Experimental results demonstrated robust performance and notable improvements across multiple gold-standard benchmark datasets, particularly in cancer tissues. Moreover, BIGraph-ST provides biologically interpretable pathway-level representations of domains, which ultimately offers a valuable tool to gain biological insights into complex tissue architectures. The source code will be publicly available upon acceptance.

  bioinformatics2026-08-05v1

• Homology-Based Variant-Effect Predictors Break Down on Cytochrome P450 Pharmacogenes

  Xu, H.; Samori, I.; Nayar, G.; Altman, R. B.

  Abstract

  Cytochrome P450 (CYP) enzymes metabolize roughly three-quarters of clinically used drugs; genetic variation in these enzymes is a leading source of interindividual differences in drug response. Predicting a variant's functional effect is therefore critical, yet the consequences of most CYP variants remain unknown. Many state-of-the-art variant-effect predictors rest on a homology-based paradigm that scores variants by evolutionary conservation -- an assumption that pharmacogenes including CYPs violate. Indeed, focusing on human CYPs, we show that homology-based models fail systematically. AlphaMissense (AM) assigns variants to its "ambiguous" class at nearly twice the proteome-wide rate across six CYPs, and within that class the scores are essentially uncorrelated with CYP2C9 DMS activity (Spearman's {rho} = 0.069); Evolutionary Scale Modeling~2 (ESM-2) shows the same pattern. We further hypothesized that non-homology-based features (sequence position, substitution chemistry, binding-site distance, and secondary structure) might help resolve the ambiguous calls, but their explanatory power is weak. Using CYP2C9 DMS activity as ground truth, we built a k-nearest-neighbors model over ESM-2 embeddings and ensembled it with AM and ESM-2 masked marginal probability, improving the ambiguous-class correlation roughly ten-fold, from {rho} = 0.069 to 0.715 (overall {rho} from 0.638 to 0.825). However, this markedly improved accuracy does not translate into agreement with clinical annotations. Drawing on evidence that a variant's effect can depend on the drug, we hypothesize that substrate identity is the key missing feature in current models, and that predicting function for these multi-substrate enzymes may require redefining function as substrate-conditioned.

  bioinformatics2026-08-05v1

• Evaluating the impact of a sample-matched reference genome on single-cell transcriptomic inferences in Plasmodium falciparum

  Almelli, T.; Dogga, S. K.; Rop, J.; Kitada, S.; Lawniczak, M.

  Abstract

  Background Plasmodium falciparum field isolates exhibit genomic variation, including copy number variation and sequence divergence. In contrast, the P. falciparum 3D7 reference genome (Pf3D7) was derived from a long-term laboratory-adapted strain and does not fully reflect the genomic variation among field isolates. The extent to which the reference genome influences RNA-seq mapping and expression inference in natural infections remains unclear. Results We generated both a reference genome and single cell RNA sequencing (scRNAseq) data from a P. falciparum-infected carrier in Mali. This new ML52 assembly was annotated using Companion with Pf3D7 as the reference. scRNAseq reads from the natural infection isolate were aligned to both the Pf3D7 genome and the isolate-specific ML52 genome, followed by locus-level alignment inspection. For most conserved genes, expression inference was concordant for both references, while genes showing reference-genome-dependent differences were investigated further. Some discrepancies were attributed to mapping artefacts or to reads aligning to unplaced genomic contigs that reflected divergent haplotypes from the co-infecting strains in the naturally infected carrier. While most multigene family loci showed concordant gene expression across both references, var genes exhibited considerable mis-mapping against 3D7 var loci as well as 2/3 of var reads not mapping at all to 3D7. Conclusion scRNAseq expression inference in P. falciparum is robust for conserved genes and most multigene families when comparing to a matched vs unmatched reference genome, but the extremely polymorphic var genes require a matched assembly in order to evaluate expression. These findings highlight the importance of the reference genome for var gene studies and should be considered when interpreting transcriptomic analyses in other organisms possessing highly variable antigenic loci.

  bioinformatics2026-08-05v1

• A structured study of cross-condition prediction of transcriptional responses to gene perturbations

  Zhu, O.; Li, J.

  Abstract

  Gene perturbation experiments coupled with transcriptomic profiling are crucial for uncovering causal gene-gene relationships, yet it remains cost-prohibitive to systematically explore perturbation responses across diverse biological conditions. As a result, in silico prediction of perturbation response has emerged as an important strategy for guiding cost-effective experimental design. Although recent methods have begun to address cross-condition perturbation prediction, it remains under-characterized across scenarios defined by whether the perturbation has been observed during training under other biological conditions. Here, we study cross-condition prediction under both seen- and unseen-perturbation scenarios. We introduce TranScouter, a lightweight encoder-decoder framework that represents perturbed genes using LLM-derived embeddings of their text summaries and represents biological conditions using transcriptomic profiles of control cells from the target condition. Across evaluated benchmarks, TranScouter performs competitively in both scenarios. We further use empirical analyses to characterize how condition-space coverage and perturbation-effect transferability shape cross-condition performance.

  bioinformatics2026-08-05v1

• A Comprehensive Database of Simulations and Meshes of Coronary Arteries from the Fame 2 Trial

  Marcinno', F.; Hinz, J.; Ando', E.; Mahendiran, T.; Buffa, A.; Deparis, S.

  Abstract

  In this work, we publish the 3D unsteady Navier-Stokes numerical simulations and meshes of coronary arteries reconstructed from invasive X-ray coronary angiograms acquired during in the Fractional Flow Reserve versus Angiography for Multivessel Evaluation 2 (FAME 2) trial. Out of the 914 clinical images, 779 vessels are successfully reconstructed and meshed. The remaining 135 vessels have been discarded since they exhibited self-intersecting geometry during the reconstruction process. The meshes are hexahedral and all of them have the same number of vertices and identical connectivity; their high quality is demonstrated using standard mesh quality indices. The simulations are performed using the Finite Element Method (FEM) with state-of-the-art coronary boundary conditions applied at the outlet. The motivation behind this effort relies from the scarcity of publicly available numerical haemodynamics data, despite the growing interest in data-driven modeling and machine learning techniques. The database is available at the link: https://doi.org/10.7910/DVN/GPCUNS

  bioinformatics2026-08-05v1

• Integrated Pangenomic and Systems Biology Analyses Reveal the Genomic Basis of Virulence and Adaptation in Bipolaris sorokiniana

  Shukla, A. K.; Kadoo, N.

  Abstract

  Bipolaris sorokiniana is a hemibiotrophic fungal pathogen responsible for foliar and root diseases of cereals, causing annual yield losses of 10-50%. The recurrent breakdown of host resistance and the emergence of fungicide-resistant pathogen populations underscore the urgent need to understand the genomic mechanisms underpinning pathogen adaptation and virulence. Here, we present the first comprehensive species-wide pangenomic and systems-level analyses of B. sorokiniana based on 19 genomes of globally distributed strains. Orthology-based analyses revealed an open pangenome comprising 16,981 orthogroups, partitioned into a conserved core genome (60.8%) and a highly dynamic accessory genome (39.2%), consisting of soft-core (8.5%), shell (19.7%), and cloud (11.0%) compartments. The core genes were predominantly associated with essential cellular and metabolic functions, while the accessory fractions were enriched in regulatory, stress-responsive, and adaptive processes. Secondary metabolite profiling identified 39-54 biosynthetic gene clusters per genome and revealed a largely conserved metabolic repertoire. Gene family evolution analyses revealed an excess of gene loss over expansion, indicating ongoing genome streamlining and lineage-specific adaptation. The core interactome comprised four densely connected functional communities governing genome maintenance, ribosome biogenesis, cellular bioenergetics, and protein translation. Collectively, this study elevates B. sorokiniana research from single-genome analyses to a population-scale analysis, providing vital insights into the evolutionary architecture of pathogenicity, adaptation, and genome diversification. These findings provide a valuable genomic resource for disease surveillance and functional characterization of virulence determinants, as well as the development of durable resistance strategies and next-generation antifungals for sustainable disease management in cereals.

  bioinformatics2026-08-05v1

• LightAlign: a lightweight pairwise aligner for memory-constrained HiFi read assembly

  Liu, J.; Zhang, J.

  Abstract

  Introduction: Current de novo genome assembly tools often demand substantial memory resources, and their execution typically relies on high-performance computing (HPC) clusters. This dependency limits their use in resource-constrained settings. Furthermore, mainstream third-generation sequencing assembly and alignment tools usually require explicit detection of overlap regions between reads, a process that often entails significant computational and storage overhead. Results: To address this issue, we developed LightAlign, a lightweight alignment tool for HiFi data that innovatively uses sequence-derived fuzzy features and reduces the peak memory usage during overlap detection. Conclusions: When combined with miniasm, LightAlign generated bacterial draft assemblies while maintaining peak memory usage below 1 GB and completed overlap generation for the tested eukaryotic datasets within 1.88 GB RAM.

  bioinformatics2026-08-05v1

• Beyond point estimates: quantifying predictive uncertainty reveals hidden dimensions of biological age acceleration and improves risk interpretation

  Wang, C.; Wu, H.; Namba, S.; Park, J. Y.; Matsuda, K.; Okada, Y.; He, Z.; Ionita-Laza, I.

  Abstract

  Biological age estimates are increasingly used to study aging, disease risk, and mortality, yet their predictive uncertainty is rarely quantified. Consequently, conventional age-gap measures can treat deviations as equally informative even when the underlying biological age predictions differ substantially in reliability. We developed a framework for uncertainty-aware biological aging that generates calibrated prediction intervals and individualized probabilities of accelerated or decelerated aging alongside point estimates. We applied this framework to the UK Biobank Pharma Proteomics Project, evaluating three composite and eleven organ-specific biological age clocks. Predictive uncertainty varied substantially both within and across clocks, revealing that apparently extreme age gaps can differ markedly in the strength of evidence supporting accelerated or decelerated aging. In particular, low-accuracy clocks, including many organ-specific clocks, provided little evidence for confidently accelerated or decelerated aging. Beyond biological age gaps, prediction-interval width was independently associated with disease risk and mortality, particularly for composite, brain, and immune clocks, suggesting that predictive uncertainty captures an additional dimension of biological aging that may reflect increased molecular heterogeneity and dysregulation associated with aging and disease. We replicated these findings in Biobank Japan and an independent clinical cohort from Stanford. By incorporating individual-specific predictive uncertainty, our framework provides a more informative characterization of biological aging and enables improved individual-level risk stratification for disease prevention and longitudinal monitoring.

  bioinformatics2026-08-05v1

• PlantNetX: A web-based transcriptomic resource integrating bulk and single-cell RNA-seq for plant functional genomics.

  Nasir, M. A.; Nawaz, S.; Faik, A.

  Abstract

  Bulk RNA sequencing and single-cell RNA sequencing provide complementary information on tissue and cell-type-specific gene expression. Bulk RNA sequencing enables the construction of gene association networks that identify co-expressed genes involved in shared pathways, whereas single-cell RNA sequencing maps their expression to cell types. However, most platforms provide access to either bulk RNA sequencing or single-cell RNA sequencing analysis, making it difficult to connect tissue-level co-expression with cell-type-specific expression. PlantNetX was developed as a web-based platform that integrates both data types. Although PlantNetX currently focuses on rice (Oryza sativa) and includes 70 quality-controlled RNA sequencing datasets comprising 1,198 sequencing libraries, together with nine single-cell RNA sequencing datasets containing more than 580,000 cells, including recently released datasets not consistently represented in existing platforms, it was designed to incorporate additional plant species, datasets, and analytical tools. PlantNetX provides Mutual Rank-based co-expression analysis, global and tissue-specific gene association networks, interactive visualization, and cell-type-specific expression summaries. The platform was validated with published examples of plant cell-wall biosynthesis and root-hair growth and retrieved gene association and expression patterns. Under standardized testing conditions, PlantNetX had a shorter mean response time than the other databases assessed. PlantNetX will support research in plant cell-wall biosynthesis, pathway discovery, functional genomics, and crop improvement.

  bioinformatics2026-08-05v1

• Anchors for Homology-Based Scaffolding

  Kaether, K.-K.; Gatter, T.; Lemke, S.; Stadler, P. F.

  Abstract

  Homology based scaffolding orders contigs based on conserved collinearity of homologous sequences across related species. Existing methods often rely on costly whole genome alignments or show limited robustness when integrating multiple references. Here, we introduce an anchor based scaffolding framework that adapts synteny anchors to efficiently infer contig order and orientation relative to one or more reference genomes. Our approach leverages precomputed, sufficiently unique anchors and their respective high confidence homology matches in a greedy approach, combining single reference to multi reference scaffolds using a maximum matching. Across simulated and real datasets, anchor based scaffolding achieves accuracy comparable to state of the art methods. Notably, the approach shows particular strengths in multi reference settings. These results demonstrate that synteny anchor based scaffolding provides an additional tool for homology based scaffolding with robust accuracy and superior performance in multi reference scenarios.

  bioinformatics2026-08-04v5

• Confidence-aware learning for transcriptome-based prediction of OXPHOS genes in Caenorhabditis elegans under incomplete functional annotation

  Zeballos - Goron, S.; Salinas, G.; Pazos Obregon, F.

  Abstract

  Assigning biological functions to genes remains a major challenge in genomics because reliable functional annotations are often scarce and unevenly distributed. Although transcriptomic datasets capture rich information about gene activity and coordination, exploiting these data for gene function prediction is difficult when only a small number of genes have high-confidence functional assignments and the remaining labels are uncertain. Here, we developed a confidence-aware learning framework that integrates complementary transcriptomic landscapes while explicitly incorporating uncertainty in functional annotations. The framework combines supervised learning from time-resolved bulk RNA-seq data with co-expression analysis of embryonic and adult single-cell transcriptomes. Supervised learning uses a two-round training scheme in which genes supported by limited functional evidence are incorporated only after an initial model has been established using high-confidence annotations. To reduce module-specific biases, we implemented an informed bagging strategy in which genes from individual functional modules are systematically withheld during training and predictions are integrated by consensus across models. We applied this framework to identify genes involved in oxidative phosphorylation (OXPHOS) in Caenorhabditis elegans. Integrating supervised and co-expression evidence prioritized a small set of high-confidence candidate genes with strong predictive performance on an independent test set. Experimental validation showed that disruption of the top-ranked candidate, ril-1, produces phenotypes consistent with impaired OXPHOS function. Our results demonstrate that transcriptomic landscapes can be systematically exploited for gene function prediction under incomplete functional annotation, providing a framework that may be transferable to other biological processes and organisms where reliable annotations remain limited.

  bioinformatics2026-08-04v4

• Longitudinal structural variant phylogenies define tumor evolution under therapeutic selection pressure in metastatic prostate cancer

  Liu, Y.; Lai, J.; Yang, Y.; Markowski, M. C.; Antonarakis, E. S.; De Marzo, A. M.; Yegnasubramanian, S.; Wood, L. D.; Sena, L. A.; Karchin, R.

  Abstract

  Therapeutic pressure shapes tumor evolution by selecting for subclones with distinct genomic architectures. In metastatic castration-resistant prostate cancer (mCRPC), structural variants (SVs) are central to this process but rarely incorporated into clonal reconstruction from bulk DNA sequencing. We describe SVCFit, a computational framework that estimates the cellular fraction of diverse SV classes from whole-genome sequencing and uses these estimates to reconstruct clonal evolutionary relationships. SVCFit explicitly models SV-type-specific breakpoint and breakend patterns and accounts for local copy-number context, enabling accurate SV cellular-fraction estimation across heterogeneous tumor genomes. On simulated datasets and in silico mixtures of metastatic prostate cancer samples, SVCFit achieves lower estimation error than SVclone, the current state-of-the-art for bulk-sequencing SV cellular-fraction estimation. Applied to longitudinal whole-genome sequencing from patients with mCRPC treated with bipolar androgen therapy, SVCFit reveals marked treatment-associated clonal reconfiguration, including contraction of highly rearranged subclones and expansion of resistant populations defined by distinct structural alterations. By enabling reconstruction of SV-defined clonal architecture from routine whole-genome sequencing, SVCFit expands the toolkit for studying tumor evolution under therapy in precision oncology. The COMBAT trial that provided these samples is registered on ClinicalTrials.gov (NCT03554317; first posted 13 June 2018).

  bioinformatics2026-08-04v4

• CPPLocPred: Subcellular Localization of Cell-Penetrating Peptides

  Bajiya, N.; Mehta, N. K.; Raghava, G. P. S.

  Abstract

  Cell-penetrating peptides (CPPs) are widely used to deliver therapeutic cargoes into cells. Although numerous computational methods have been developed for identifying CPPs and several predictors are available for protein subcellular localization, no method has been developed to predict the subcellular localization of CPPs. Here, we present CPPLocPred, a hierarchical machine-learning (ML) framework that predicts CPPs and their subcellular localization. In the first stage, we developed ML models to identify CPPs, achieving an AUC of 0.953 with an MCC of 0.7842 on an independent set, exhibiting performance equivalent to or better than existing state-of-the-art methods. In the second stage, we developed a method for predicting the subcellular localization of CPPs. Subcellular localization methods were trained (80% data using five-fold cross-validation) and validated (20% data) on experimentally validated CPPs for 663 Cytoplasm, 287 Nucleus, 57 Mitochondria, 186 Endo_lysosome, and 328 Others. Our primary analysis revealed that Mitochondrial and Nuclear associated CPPs are abundant in positively charged arginine- and lysine-rich patterns, whereas Endo_lysosomal CPPs preferentially comprise glycine-, proline-, and cysteine-rich motifs. We used a wide range of traditional peptide features, along with the embedding of protein language models, to develop ML models. Among all evaluated models, the CatBoost-based subcellular localization models with Distance Distribution of Residues (DDR) achieved AUCs of 0.814, 0.775, 0.970, 0.782, and 0.798 for Cytoplasm, Nucleus, Mitochondria, Endo_lysosome, and Others, respectively, on validation dataset. We developed CPPLocPred, which offers a practical platform for functional annotation and rational design of localization-specific CPPs for therapeutic applications (https://webs.iiitd.edu.in/raghava/cpplocpred/). HighlightsO_LIPrediction and subcellular localization of cell-penetrating peptides. C_LIO_LILocalization of CPPs depends on their amino acid and dipeptide composition. C_LIO_LIBest feature for subcellular localization was Distance Distribution of Residues. C_LIO_LICatBoost model achieved the highest performance for subcellular localization. C_LIO_LIA web server and standalone software to facilitate its use by the scientific community. C_LI

  bioinformatics2026-08-04v2

• DIVAS: an R package for identifying shared and individual variations of multiomics data

  Sun, Y.; Marron, J. S.; Le Cao, K.-A.; Mao, J.

  Abstract

  Motivation: Multiomics data integration aims to identify biological patterns shared across molecular modalities. Most existing methods detect either jointly shared variation, across all modalities, or individual variation, unique to a single modality, but overlook partially shared variation, shared by only a subset of modalities. This is a critical limitation, because many biological mechanisms manifest in some but not all molecular modalities. Results: We present an open-source R package implementing DIVAS, a framework for systematically identifying jointly shared, partially shared and individual variations across multiple data types. DIVAS combines angle-based subspace analysis with inference through rotational bootstrap, hierarchically searching all combinations of modalities to decompose multiomics data into interpretable components with scores and loadings. In simulations with a known sharing structure, DIVAS recovered every component across a wide range of noise levels, whereas AJIVE and MOFA+ did not. Applied to multi-modal COVID-19 data, it reveals partially shared immune and metabolic dysregulation patterns underpinning disease severity that conventional approaches would miss. Availability and implementation: DIVAS is available as an R package on GitHub (https://github.com/ByronSyun/DIVAS). A step-by-step vignette is available on GitHub (https://byronsyun.github.io/DIVAS_COVID19_CaseStudy/).

  bioinformatics2026-08-04v2

• Variational inference in coupled models of amino acid substitution

  Large, A. L.; Holmes, I. H.

  Abstract

  We investigate the use of Expectation--Maximization (EM) and variational Bayes for inferring rates and interactions under models of molecular coevolution. We first review EM theory for continuous-time Markov chains (CTMCs) and develop it for coevolutionary models, exploiting exchangeability and reversibility symmetries to constrain the parameter dimension. We fit several paired amino-acid coevolutionary models to structural alignments and compare the results to previous work. Our richest model trained on pooled coevolutionary data has explanatory power comparable to CherryML's Q2 matrix (also trained on pooled data), with one quarter the parameters. However, we observe that aggregation of training data can lead to a form of Simpson's Paradox: a mixture model, whose components are parameter-efficient continuous-time Bayes networks (CTBNs), resolves signals that wash out when a single model tries to capture everything. These signals include both correlated and anticorrelated hydropathy and volume-packing in coevolving amino-acid pairs, as well as the anticorrelated acid/base compensation that was detected by CherryML's Q2. We next present a closed-form evidence lower bound (ELBO) for CTBNs using EM statistics. Compared to the state of the art in variational modeling of CTBNs, the Euler-Lagrange equations derived by Cohn et al (JMLR, 2010), our closed-form ELBO is competitive in accuracy, considerably more efficient, simpler, and more stable. We conclude by describing a covariant indel model: a Dirichlet process selecting coevolving sites within TKF92, yielding an Infinite Pair HMM over alignments and structures. This model slightly outperforms TKF92 on a structural-alignment benchmark. Code and data are at https://tkfdp.net/.

  bioinformatics2026-08-04v2

• Interpreting Protein Language Models: high attention sites predict functional regions

  Pribus, S. J.; Altman, R. B.; Nayar, G.

  Abstract

  Computational proteomics has revolutionized biomedical research, guiding targeted experimental exploration to accelerate protein-based mechanistic discovery. Protein Language Models (PLMs) enable scalable, resource-efficient study; through large-scale training on only primary protein sequences, PLMs generate vector representations of protein structure that have been shown to capture biochemical, evolutionary, and structural properties. A core component of PLMs is the attention mechanism, which specifically captures long-range interactions across a protein sequence in attention matrices. Using the Evolutionary Scale Modelling 2 (ESM-2) PLM, we previously developed a novel method to identify "High Attention" (HA) sites. HA sites are specific residues that ESM-2 assigns the most attention to early during encoding. Here, we further characterize these HA sites across structural and functional metrics. Using unsupervised clustering, we find HA sites can be categorized as "structural core", "structural pathogenic", "core pathogenic", or "low-confidence". We further use AlphaMissense pathogenicity predictions and the pan-cancer analysis of whole genomes (PCAWG)-labeled pathogenic variant positions to show that HA sites predict protein regions with high pathogenic risk. Finally, we explore the utility of HA sites for suggesting candidate binding sites, identifying multiple cancer protein examples where HA sites identified regions with previously undiscovered high interaction likelihood and thus potential therapeutic utility. Our work demonstrates the biological interpretability of PLM representations and offers a valuable method to prioritize functionally relevant protein residues for targeted biomedical research.

  bioinformatics2026-08-04v1

• The Fontan EV Score: A Circulating Extracellular Vesicle-Based Risk Stratification Tool for Fontan-Associated Liver Disease

  Takaesu, F.; Li, X.; Kievert, J.; Zhou, A.; Kemper, S.; Yuhara, S.; Hussain, S.; Watanabe, T.; Matsuda, J.; Taha, F.; Morrison, A.; Nelson, K.; Zucco, J.; Naguib, A.; McKee, C.; Hill, J.; Carrillo, S. A.; Breuer, C. K.; Kelly, J. M.; Brigstock, D.; Davis, M.

  Abstract

  Background: Fontan-associated liver disease (FALD) is a universal complication of the Fontan palliation characterized by chronic congestion and progressive hepatic fibrosis. Current diagnostics rely on invasive biopsies or non-specific biomarkers that fail to capture early fibrogenesis, creating a need for non-invasive biomarkers to stratify disease severity. Methods: We utilized an ovine Fontan model (n = 19) to investigate circulating serum extracellular vesicles (sEVs) as reporters of hepatic pathology. Longitudinal samples paired with liver elastography were collected, and sEVs were subjected to multi-omic profiling including small RNA sequencing and proteomics. Regularized regression was used to identify transcriptomic predictors, which were integrated with time post-surgery into an ordinal logistic regression framework to construct the Fontan EV Score (FES). Model performance was evaluated on a held-out test cohort and benchmarked against established fibrosis indices. To validate the biological relevance of the FES , TGF-{beta}-treated human liver organoids were generated and scored miRNA expression was assessed. Results: The sEV proteome exhibited robust separation by surgical physiology, while the small RNA cargo was primarily stratified by fibrotic status. Bioinformatics confirmed a high hepatic origin for these transcripts and identified enrichment of inflammatory pathways including Toll-like receptor and Interleukin-17 cascades in fibrotic subjects. The FES, incorporating time post-surgery and eleven small RNA biomarkers, demonstrated high predictive accuracy in the independent testing cohort with an AUC of 0.876 for moderate and 0.963 for severe fibrosis, substantially outperforming APRI (AUC = 0.618) and FIB-4 (AUC = 0.731). In human liver organoids, several scoring miRNAs, including miR-125a-5p and miR-193b-5p, were directionally responsive to profibrotic stimulation. Conclusions: Circulating sEVs carry a liver-associated cargo that can be leveraged for the non-invasive prediction of FALD severity. The FES provides a biologically validated scoring system that substantially outperforms existing serological indices and offers a new avenue for early detection and risk stratification of FALD.

  bioinformatics2026-08-04v1

• MAXWELL: Calibrating the probabilistic outputs of protein language models to the mutation-induced stability change landscape

  Li, M.; Cheng, X.; Jiang, F.; Hong, L.; Yu, Y.

  Abstract

  Designing mutations that enhance protein stability is a central goal in protein engineering. However, experimentally screening large numbers of candidate mutations is costly and time-consuming, creating a strong need for computational methods that can identify potentially stabilizing mutations. Among these approaches, protein language models are particularly promising because they learn context-dependent amino acid preferences from large-scale sequence and structure datasets. Nevertheless, most existing stability prediction methods use these models primarily as feature extractors and do not fully exploit the amino acid probability distributions they encode. Here, we introduce MAXWELL (Matrix-wise Landscape Learning), a novel post-training method that calibrates the probabilistic outputs learned by protein language models during pretraining to generate mutational landscapes that quantify the effects of individual amino acid substitutions on protein stability. When applied to ProteinMPNN, MAXWELL yields a state-of-the-art predictor of the effects of protein mutations on stability, outperforming ThermoMPNN and other representative methods on a curated benchmark of experimentally measured stability changes. We next applied MAXWELL to the design of ten single-point mutations in the DhaA dehalogenase, seven of which (70%) increased thermal stability. Among them, G171W showed the largest improvement, with a measured {Delta}T of 4.91 . These experimental results establish MAXWELL as a novel post-training strategy for protein language models and a practical framework for designing stabilizing mutations.

  bioinformatics2026-08-04v1

• Pretrained deep-learning ITS classifiers read the flanking regions, not the ITS2 barcode, and so fail on the amplicon that environmental fungal surveys sequence

  O'Brien, A.; Parada, P.

  Abstract

  Deep-learning classifiers for the fungal internal transcribed spacer (ITS) report accuracies above 90% and are increasingly proposed for environmental metabarcoding. We benchmark two pretrained models, a convolutional network and a transformer sharing a training corpus of 5.23M sequences, against two established k-mer methods on 5,222 identical queries, one per genus, evaluated on both full-length ITS and the ITS2 subregion that dominates environmental sequencing. The design favours the classifiers: queries are drawn from the same public dataset they were trained on and stratified by whether a query's genus lies in their own label space, recovered from the distributed models, while the reference the k-mer methods consult mirrors that label space and excludes the queries themselves. Even so, on full-length ITS both classifiers are outperformed by both classical methods at every rank and in both strata: SINTAX recovers the correct family for 92.0% of seen-genus and 70.3% of novel-genus queries and best-hit alignment against a 56,327-sequence reference for 92.3% and 67.2%, against 77.9% and 57.7% for the transformer and 76.5% and 54.3% for the convolutional network. A hierarchical logistic regression on k-mer counts, fitted in ten minutes to 1.07% of the MycoAI training corpus, also exceeds both and places novel genera better than either search method, and refitted on ITS2 it recovers 89.2% of seen-genus families on that amplicon against 88.6% for best-hit alignment, so neither learned classification nor the amplicon is what fails. Restricting the identical records to ITS2 costs the k-mer methods 3.5 and 3.7 percentage points of seen-genus family accuracy but costs the classifiers 49.6 and 58.0, reducing them to 28.3% and 18.5%. An ablation identifies the cause. Grafting each query's unaltered ITS2 between the flanking regions of a donor record from a different phylum returns the donor's family for 34.4% of queries against the query's own for 4.2% in the convolutional model, and 63.7% against 0.8% in the transformer, from a baseline of 0.1% where no donor sequence is present. The models therefore read taxonomy principally from the flanking regions rather than from the ITS2 barcode, which explains the collapse and predicts the same failure for any subregion amplicon. Compounding this, on ITS2 the classifiers' output probability all but ceases to separate novel from known genera (AUROC 0.541 and 0.503, the latter at chance, against 0.866 for alignment identity and 0.785 for the SINTAX bootstrap), so the failure is not detectable from the models' own output. We recommend that reported accuracies for such models specify the amplicon region of the evaluation, state the length distribution of the training corpus, and include a same-query classical baseline.

  bioinformatics2026-08-04v1

• A calibrated novelty flag for fungal ITS metabarcoding: choosing the error rate at which sequences are declared new

  O'Brien, A.; Parada, P.

  Abstract

  Environmental fungal surveys routinely recover internal transcribed spacer (ITS) sequences that cannot be assigned at fine taxonomic ranks, the so-called fungal "dark matter," and in practice such sequences are set aside by thresholding a similarity or confidence score at a conventional value. Those conventions do abstain, but the error rate a given threshold implies is neither stated nor selectable, and a threshold defined for one kind of score does not transfer to another. We present a conformal novelty flag that supplies what is missing: each query receives a p-value with a distribution-free guarantee that the rate of falsely declaring a known sequence novel is bounded by a user-chosen alpha. On a leave-one-genus-out benchmark built from UNITE the flag holds its nominal rate across two orders of magnitude in alpha, so the practitioner selects an operating point on a stated frontier rather than inheriting one: at alpha=0.05 it fires on 4.3% of known-genus queries and recovers 19.2% of genuinely novel genera, and at alpha=0.20 on 19.7% and 50.9%. It calls 4,996 of 5,000 non-fungal SSU outgroup sequences novel, anchoring the far end of a monotonic known-to-outgroup gradient. The guarantee is score-agnostic in practice and not only in principle: applied to alignment identity, a k-mer bootstrap consensus and two neural classifiers' output probabilities, across two amplicon regions and a sevenfold change in reference size, all eleven calibrations fell within 1.1 percentage points of nominal. Detection over those same eleven settings ranged from 5.2% to 53.3%, a tenfold spread under an essentially constant error rate, which makes explicit that the guarantee is on the error rate and not on power: a well-calibrated flag built on an uninformative score is valid and useless, and two of our settings are exactly that. Applied to a soil fungal dataset the flag identifies 20.7% of amplicon sequence variants as novel at a controlled 5% error rate, falling to 16.3% under an abundance filter. Half of those recur near-identically among the unnamed environmental variants of GlobalFungi while fewer than one in ten matches a named species hypothesis, a sixfold skew towards the uncatalogued against 1.9-fold for the sequences the flag passes: present in soils worldwide and named in almost none of them. The flag turns an arbitrary cutoff into a decision with a stated error rate, and in doing so converts dark matter from a residue into a set of prioritizable targets. Source code and the benchmark harness are available under the MIT license at https://github.com/ayobi/its-novelty.

  bioinformatics2026-08-04v1

• Scientific computing in the age of agentic AI: an exploratory field report

  Li, J.; Rubinsteyn, A.; Feldman, S.; O'Donnell, T.; Ferguson, J. M.; Patro, R.; Driver, I.; Ewels, P. A.; Krueger, F.; Angerer, P.; Gold, I.; Manning, J.; Heumos, L.; Ahangari, M.; Goyal, V.; Masoudi, H.; Pedersen, B.; Bai, A.; Li, H.; Shringarpure, S.; Ho, A.

  Abstract

  Scientific computing has become a central component of modern scientific discovery. Yet many computational tools are developed by small, specialized teams under incentives that encourage the release of rapidly prototyped tooling without commensurate attention to engineering concerns, including performance and maintainability. These gaps are particularly visible in the life sciences, where the advent of high-throughput sequencing and molecular profiling has made the production and processing of datasets routine at scales that strain reliability and cost. Recently, LLM-based agents have become increasingly capable, with publicly available systems possessing both significant domain knowledge in many scientific fields and the ability to autonomously operate over complex and specialized codebases in pursuit of well-defined goals. Together, these developments create a practical opportunity for scientific computing. Many of the persistent weaknesses of the scientific computing ecosystem stem from technical debt and a shortage of sustained engineering labor and expertise. Here, we examine coding agents as a potential way to address these weaknesses: we present an exploratory field report of eight early case studies in the application of LLM agents to scientific computing across a range of project scopes, from lightweight maintenance tasks to full performance-oriented rewrites of scientific libraries, with a focus on the life sciences. Each of these case studies is accompanied by reflections from the individual or group responsible for the work, including lessons from the process. Overall, we find that the use of coding agents in scientific computing holds great promise for accelerating scientific research and increasing the reliability of critical systems, but that outstanding concerns remain, including responsibility and ownership for such projects, and we suggest collaboration and stewardship with existing maintainers when feasible.

  bioinformatics2026-08-04v1

• Evolutionary Design of Membrane-Lytic Antimicrobial Peptides with Mixture of Experts

  Dong, R.; Song, C.

  Abstract

  Antimicrobial peptides (AMPs) hold great promise in combating drug-resistant pathogens, yet current design models do not explicitly account for their primary mode of action, membrane disruption. To address this gap, we propose AMPainterV2, a model that designs membrane-lytic AMPs using generative adversarial imitation learning. AMPainterV2 integrates a Mixture-of-Experts (MoE) policy network to support insertion, deletion, and mutation operations, thereby expanding the evolutionary design space. In vitro experiments demonstrate that all ten top-ranked peptides evolved from random sequences are active membrane-lytic AMPs, with the best candidate achieving a mean minimal inhibitory concentration (MIC) of 1.5 M and superior membrane-lytic activity compared to polymyxin B. Particularly, a deletion-only version of AMPainterV2 enables membrane-lytic AMP miniaturization, yielding six miniaturized peptides with comparable or improved activity and an average 36% length reduction. Collectively, AMPainterV2 advances mechanism-driven AMP design and offers a valuable tool for cost-effective AMP development.

  bioinformatics2026-08-04v1

• ComBatFamQC: Streamlining Interactive Batch-Effect Diagnostics and Harmonization for Neuroimaging Data in R

  Ren, Z.; Horwath, E.; Wen, S.; Melhem, R.; Anderson, J. K.; Johnson, W. E.; Shinohara, R. T.; Chen, A. A.; Shou, H.

  Abstract

  As multisite and multi-study data aggregation becomes increasingly common for improving statistical power and sample diversity, robust harmonization methods are needed to address biases introduced by batch variation, particularly in neuroimaging research. Although a variety of harmonization approaches are available, the lack of systematic guidance for diagnosing batch effects and selecting appropriate methods remains a major challenge. To address this gap, we introduce ComBatFamQC, a comprehensive R package designed to streamline batch-effect diagnosis, harmonization, and post-harmonization analysis. ComBatFamQC integrates a user-friendly Shiny app for interactive batch-effect diagnostics, state-of-the-art harmonization methods from the ComBat family, including ComBat, longitudinal ComBat, ComBat-GAM, and CovBat, and tools for downstream analysis after harmonization. The package provides qualitative visualizations, statistical tests for batch-effect assessment, and a consistent interface that supports both in-sample and out-of-sample harmonization through the Shiny app, the R console, or the command line. In addition, it includes functions for post-harmonization analyses to facilitate downstream modeling. Its modular design also supports the systematic incorporation of future harmonization methods and expanded downstream analysis capabilities.

  bioinformatics2026-08-04v1

• Spatial Transcriptomics of Glioblastoma Defines Spatially Anchored Transcriptional Activity Informing Therapeutic Vulnerability and Resistance

  Ravikumar, V.; Maddox, A.; Kulkarni, R.; Hara, T.; Rao, A.; Al-Holou, W.

  Abstract

  Intratumoral heterogeneity is thought to hinder targeted therapy in glioblastoma (GBM), but the extent of variation in known targets and tumor cell-intrinsic and extrinsic regulatory mechanisms within tissue remains unclear. In this work, we use Visium spatial transcriptomics to profile 44 tumor slides and 128,176 niche-annotated ST spots, including 14 new slides from 7 GBM specimens, along with spatial annotation from the Ivy Glioblastoma Atlas Project. First, we characterize the differential expression of AVB3, EGFR, VEGF, and PDGFRA pathway activities across space, elucidating the limited success of individual targeted therapies. Second, we identify transcription factor modules driving intratumoral heterogeneity and characterize a novel transcription factor module at the interface between core tumor and the invasive edge. Third, we identify cell type-specific receptor-ligand signaling enriched in specific spatial niches and relate these to specific molecular program alterations. Finally, we identify and characterize transcriptional pathways and their activity across spatial niches through transcription network analysis and demonstrate how these pathways can inform a combined therapeutic approach. Together, through novel insights into the spatial patterning of transcription factor regulation, cellular interactions, and biological pathway activity, our work informs rational combination therapies targeting spatial niche specific vulnerabilities.

  bioinformatics2026-08-03v3

• BioGraphX-RNA: A Universal Physicochemical Graph Encoding for Interpretable RNA Subcellular Localization Prediction

  Saeed, A.; Abbas, W.

  Abstract

  RNA subcellular localization is a critical determinant of cellular function. However, current computational approaches often operate as ""black boxes", overlooking the complex interplay among sequence, structure, and physicochemical interactions that govern RNA localization. Building upon the BioGraphX framework originally developed for proteins, we introduce BioGraphX-RNA, a universal physicochemical graph-encoding framework that provides a structure-informed encoding by translating primary nucleotide sequences into multiscale interaction graphs using explicit biophysical rules. When combined with frozen RiNALMo embeddings via an interpretable gated fusion layer, BioGraphX-RNA achieves competitive performance with DeepLocRNA and uniquely quantifies the relative contribution of sequence versus structure for each RNA. On human datasets, the gated fusion model attains macro AUROC values of 0.7575 +/-0.0054 (mRNA), 0.9228 +/- 0.0137(miRNA), and 0.5600 +/- 0.0191 (lncRNA). For miRNA, the graph only model alone reaches 0.9396 +/-0.0045, outperforming both the RiNALMo language model and a RNAfold partition function graph (0.9139 +/- 0.0138), validating the structure;informed proxy hypothesis. In a blind cross-species prediction task on mouse data, the model shows limited zeroo shot transfer, indicating that biophysical graph features do not improve cross species generalization. Gating analysis reveals RNA type-specific modality reliance, with miRNA exhibiting a near equilibrium balance between sequence and structure. SHAP based interpretation suggests potential correlates such as patterned GC content for nuclear retention and structural accessibility for exosome targeting. These advances are achieved with only 2.05 million trainable parameters, aligning with Green AI principles. BioGraphX-RNA demonstrates that explicitly integrating biophysical constraints into graph-based encodings enables accurate and interpretable predictions for structured RNAs, advancing structure-aware RNA biology and laying a foundation for precision medicine.

  bioinformatics2026-08-03v3

• RKMR: A Rapid Kernel Machine Regression Framework for Optimal Marker Detection in Spatial Omics Data

  Seal, S.; Chakraborty, A.; Mattila, C.; Rubinstein, M.; Angel, P.; Ghosh, D.; Chung, D.; Neelon, B.

  Abstract

  High-throughput spatial omics technologies enable molecular profiling within intact tissue architecture, yet identifying concise, predictive, and biologically interpretable marker panels for cell types, tissue domains, and disease-associated tissue classes remains challenging. This limitation hinders the development of actionable panels for targeted validation and downstream translation. Existing pipelines rely largely on univariate differential-expression analyses, which ignore joint molecular structure and provide limited predictive insight. Multivariate machine-learning methods, including random forest, XGBoost, elastic net, and specialized single-cell panel-selection approaches, can capture predictive patterns but typically lack explicit spatial modeling and probabilistic feature selection, relying instead on model-specific importance scores or user-specified panel sizes. We develop rapid kernel machine regression (RKMR), a scalable framework for spatial-omics marker discovery that integrates nonlinear kernel modeling, spike-and-slab variable selection, and spatial dependence. RKMR uses automatic relevance determination (ARD) kernels and sparsity-inducing priors to capture nonlinear marker-outcome relationships and implicit feature interactions while producing approximate posterior inclusion probabilities (PIPs) that quantify model-based uncertainty in feature inclusion. To scale inference to large spatial datasets, RKMR combines low-rank kernel approximations with stochastic variational optimization. In simulations, RKMR consistently achieves higher AUPRC than competing methods across a range of molecular-signal and spatial-effect settings. Across spatial transcriptomics and scRNA-seq datasets, RKMR identifies parsimonious marker sets that recover reported cell-type signatures and reproducible tissue-layer markers. These results establish RKMR as a scalable and uncertainty-aware framework for translating high-dimensional spatial omics data into robust, experimentally actionable marker panels.

  bioinformatics2026-08-03v3

• A Systematic Comparison of Single-Cell Perturbation Response Prediction Models

  Li, L.; You, Y.; Fu, Y.; Liao, W.; Fan, X.; Lu, S.; Cao, Y.; Li, B.; Ren, W.; Kong, J.; Zheng, S.; Chen, J.; Liu, X.; Tian, L.

  Abstract

  Predicting single-cell transcriptional responses to perturbations is central to dissecting gene regulation and accelerating therapeutic design, yet the field lacks a rigorous, task-spanning assessment of model behavior. We present a large-scale benchmark of 13 representative methods and baselines across 25 datasets spanning diverse perturbation modalities and species, including two primary immune-cell drug-response resources. We evaluated three core tasks-generalization to unseen single-gene perturbations, prediction of combinatorial interactions, and transfer across cell types-using 24 metrics covering expression-level accuracy, relative changes, differential expression recovery, and distributional similarity.Across tasks, performance depended strongly on perturbation effect size and evaluation perspective: expression-level agreement was highest for small-effect perturbations resembling controls, whereas delta- and DE-based metrics improved with larger effects, providing clearer signals. Models shared a conservative bias, with fine-tuned foundation models compressing variance and underestimating synergistic effects in combinations. PerturbNet showed superior recovery of DE signatures in Tasks 1 and 2, while no method consistently generalized across cell types in Task 3, where biological consistency dominated outcomes.This benchmark establishes current methodological limits, clarifies that different metrics probe distinct biological signals rather than redundant summaries of the same prediction problem, and provides a foundation for developing virtual-cell models that more faithfully capture heterogeneous perturbation responses.

  bioinformatics2026-08-03v3

• Explainable Generative AI Uncovers a Molecular Continuum in Medulloblastoma with Implications for Rare Cancer Subtyping and Treatment Equity

  Prol-Castelo, G.; Tejada-Lapuerta, A.; Urda-Garcia, B.; Nunez-Carpintero, I.; Garcia-Verellen, E.; Montagud, A.; Valencia, A.; Cirillo, D.

  Abstract

  Medulloblastoma is a childhood brain tumor traditionally classified into four molecular subgroups. Recent evidence suggests that Groups 3 and 4 represent a biological continuum rather than distinct entities, a paradigm shift with significant implications for understanding disease biology and treatment strategies. Nevertheless, assessing this hypothesis is challenging mainly due to data scarcity. In this study, we analyze the largest available transcriptomics dataset to provide compelling evidence for the existence of an intermediate subgroup between Groups 3 and 4, characterized by distinct molecular features. To overcome limitations posed by data scarcity, we employ synthetic data generation using a Variational Autoencoder and apply explainability techniques to identify key relationships between gene expression and disease subgroups. Furthermore, by incorporating Machine Learning Fairness approaches, we demonstrate that overlooking this intermediate subgroup can result in treatment disparities. Our findings are further supported by both existing and newly proposed studies using diverse datasets and methodologies, including graph-based analyses and multi-scale simulations, underscoring the robustness and reproducibility of our results. This study demonstrates the potential of synthetic data generation to refine rare disease subtyping and advance our understanding of the underlying biological mechanisms.

  bioinformatics2026-08-03v2

• Do Geometric Outliers Identify Important Genes in Single-Cell Foundation Models?

  Whalley, J. P.

  Abstract

  Single-cell foundation models (scFMs) produce gene embeddings that are increasingly interpreted biologically, but it is unclear whether different models organise gene space consistently, or whether geometric extremes identify important genes. The same four-metric screen was applied to Geneformer, scGPT and scFoundation, which differ in architecture, training objective and expression encoding. The models agreed weakly on individual outlier genes but showed structured class-level convergence: ribosomal enrichment occurred in all three, although weaker and caller-dependent in scFoundation, while strong mitochondrial enrichment was confined to Geneformer and scGPT. Gene-level agreement was itself uneven, with Geneformer and scGPT overlapping above chance and scGPT and scFoundation not, and caller stability differed between models. Most outliers were not extreme in ESM-2 protein-sequence space, indicating principally model-specific rather than protein-sequence geometry. These patterns did not translate into the tested notions of gene importance: deleting the highest-anomaly Geneformer genes did not impair cell-type annotation beyond matched controls, and covariate-adjusted outlier status was not associated with ClinVar membership in any model. Gene-embedding geometry therefore characterises model-specific structure, but does not provide standalone evidence of downstream leverage or disease relevance.

  bioinformatics2026-08-03v2

• crispAIPE: Probabilistic Modelling of Prime Editing Variant Correction Efficiency

  Ozden, F.; Lu, P.; Minary, P.

  Abstract

  Prime editing installs precise substitutions, insertions, and deletions without double-strand breaks or donor templates, but the efficiency of an individual pegRNA design is hard to predict in advance, and existing tools give only point estimates, leaving designers unable to judge which predictions to trust. We present crispAIPE, a transformer-based probabilistic framework that quantifies per pegRNA-target pair uncertainty by modelling the three competing prime-editing outcomes on the 2-simplex with a Dirichlet likelihood and pairing the posterior with split-conformal highest-density regions calibrated on a held-out fold, giving finite-sample coverage guarantees on the simplex without requiring the Dirichlet model itself to be well-calibrated. Trained on 92,423 PRIDICT Library-1 pegRNAs under mutation-level target-disjoint splitting, crispAIPE attains Spearman rho = 0.835, 0.843, and 0.693 on the edited, unedited, and indel fractions (Pearson r = 0.845, 0.855, 0.659), and its conformal regions match nominal coverage where region-construction baselines do not. pegRNA architecture and edit context, in particular GC content of the mutated reverse transcription template (RTT) and edit size for deletions, are associated with prediction uncertainty. Reusing the Library-1 calibration quantile unchanged on PRIDICT Library-2, the conformal region area generalises as an actionable filter for cross-cell-type design, identifying the pegRNAs whose Library-1 predictions transfer best while small-sample head fine-tuning further improves accuracy. Tool and trained models: https://github.com/furkanozdenn/pe-uncert.

  bioinformatics2026-08-03v2

• PG-LLM: Benchmarking General-Purpose Language Models for Protein Variant Ranking

  Arora, R. K.; Chen, L. T.; Du, M.; Marks, D.; Church, G.

  Abstract

  eneral-purpose language models are being increasingly utilized in protein-design workflows, yet their ability to evaluate variant effects remains unclear. To answer this question, we introduce PG-LLM, a benchmark built on ProteinGym to evaluate general-purpose language models on 217 protein-variant prioritization tasks. Each task follows the same format: a language model receives a wild-type protein sequence, an assay description, and is tasked with ranking 50 mutant sequences by fitness without access to multiple-sequence alignments or protein structures. We evaluate thirteen language models and rescore 95 published protein predictors on the same candidate sets with the same evaluation metric. Claude Opus 5 leads the primary leaderboard with a Spearman correlation of {rho} = 0.406, narrowly ahead of GPT-5.6 Sol at 0.402. However, GPT-5.6 Sol scores higher than Opus 5 when the two models are compared only on assays scored by both. Opus 5 outperforms 49 of 95 published protein predictors, including 41 of 46 sequence-only methods, and approaches ESM2-650M at {rho} = 0.411, but remains below the leading predictor VenusREM at {rho} = 0.523. Variant-ranking performance improves with test-time compute across GPT, Claude, and Gemini models, but the gains taper before closing the gap to specialist protein predictors. Unlike sequence-only predictors, which perform better on proteins with deeper evolutionary alignments, LLM accuracy changes little across alignment-depth. To address contamination risk, we additionally source 59 DMS assays from 19 studies whose scores first became public after January 2026. On this set, we evaluate three LLMs and seven biomolecular baselines, observing performance trends similar to those on the main PG-LLM benchmark. PG-LLM shows that tool-free language models capture substantial protein-variant signal and already outperform many established sequence-based predictors. These results establish the emerging capability of language models as biomolecular reasoners while defining the remaining headroom for their reliable use in variant-prioritization workflows.

  bioinformatics2026-08-03v2

• vartracker: an end-to-end tool for pathogen longitudinal variant analysis and visualisation

  Foster, C. S. P.; Rawlinson, W. D.

  Abstract

  Longitudinal sequencing can reveal fine-grained pathogen evolution during acute and chronic infections and inform public health responses. However, integrating ordered pathogen genomic data into a coherent evolutionary and clinical framework can be tedious and error-prone. We present vartracker, an open-source tool for longitudinal pathogen variant analysis and visualisation. Given an ordered sample manifest, vartracker supports three entry points: raw sequence reads, reference-aligned BAM files, or user-supplied VCF and coverage inputs. Raw-read and BAM inputs are processed through an integrated Snakemake workflow, whereas VCF mode starts from precomputed files. Variants are normalised and annotated relative to a reference genome, tracked across timepoints, and classified as original or newly emerging and as transient or persistent. Inferred amino acid changes are reported, and for SARS-CoV-2 analyses, relevant published literature for key mutations can be automatically linked through a functional database. vartracker outputs a schema-documented results table, provenance metadata for reproducibility, publication-quality static figures, and an interactive heatmap for data exploration. Although packaged with SARS-CoV-2 reference assets and initially developed for SARS-CoV-2 datasets, vartracker is pathogen-agnostic when appropriate reference data are supplied. We demonstrate its utility using SARS-CoV-2 and respiratory syncytial virus A (RSV-A) datasets. vartracker is freely available through GitHub, PyPI and Bioconda.

  bioinformatics2026-08-03v2

• What limits local ancestry inference at low divergence: a feasibility threshold, a metric that conceals failure, and a deficit of input more than architecture

  Tian, Q.

  Abstract

  Local ancestry inference assigns each position along an admixed chromosome to a source population, underpinning admixture mapping, ancestry-specific association testing and admixture dating. Validation is almost exclusively on continentally divergent sources (Hudson's FST {approx} 0.1) and coalescent simulations; we examine both restrictions. Across FST from 0.0022 to 0.243 we compare five methods--two likelihood baselines, RFMix, FLARE and a dilated convolutional network--on identical sites with exact ground truth, and on 11 real 1000 Genomes pairs. Three findings follow. First, a feasibility floor: at FST = 0.0022 no method exceeds 0.575, and CHB/CHS at FST = 0.00042 yields at best 0.551. Pairs motivating fine-scale analysis, such as northern versus southern Han, fall below it. Second, per-site accuracy conceals a failure of tract structure: the most accurate method per site produces 78.8x too many tracts, implying an admixture time 61.2x too old, which Viterbi decoding removes at no cost to accuracy (+0.0002). Third, the simulated lead does not survive real data, and the deficit is one of input more than of the architectures we varied: attention, state-space layers, capacity, objective and self-supervised pretraining each move accuracy by at most 0.006, while supplying the haplotype information the released tools receive recovers +0.031 on 8 of 8 pairs below FST = 0.04 and nothing above it--necessary but not sufficient, since the network still trails on 10 of 11 pairs. Two quantities usually held fixed matter more than architecture: the statistic summarising reference matching, and reference panel size, which no method is near saturating.

  bioinformatics2026-08-03v1

• STAR Suite: Transcriptomics processing in a single binary through AI-assisted development

  Hung, L.-H.; Baker, D.; Flynn, B.; Huangfu, D.; Luo, R.; Robson, P.; Zhou, T.; Yeung, K. Y.

  Abstract

  Single-cell transcriptomic assays - pooled CRISPR/Perturb-seq screens, clonal lineage tracing, and 10x Flex fixed RNA profiling - are central to modern biology, yet their processing has depended on Cell Ranger, a proprietary tool restricted to 10x products, with no open end-to-end pipeline for Perturb-seq or Flex. We present STAR Suite, which extends the STAR aligner into a single executable that processes bulk RNA-seq, scRNA-seq, Perturb-seq, 10x Flex, and SLAM-seq in one command, with no added dependencies. For single-cell assays it reproduces Cell Ranger 9.0.1 (gene-level Pearson 0.99-1.0) while running 3.6- to 5.7-fold faster, via novel exact-matching and alignment-validated hashing algorithms. Served to both AI agents and human operators through one validated interface, STAR Suite is the production standard for the NIH MorPhiC consortium; its 132,226 lines of C/C++ were built under a human-directed, AI-implemented workflow. Source code, workflow recipes, and per-run provenance are released under the MIT license

  bioinformatics2026-08-02v6

• MGMG: Cell Morphology-Guided Molecule Generation for Drug Discovery

  Tang, Q.; Ding, D.; Yuan, X.; Seabra, G.; Ramdhan, P. A.; Liu, C.-Y.; Thai, M. T.; Li, C.; Luesch, H.; Li, Y.

  Abstract

  Designing novel molecules with desired bioactivity remains a fundamental challenge in drug discovery. Most molecular design methods follow target-based drug discovery paradigms that rely on well-defined drug targets, thereby limiting their applicability to diseases lacking known targets or reference compounds. Here we introduce Morphology-Guided Molecule Generation (MGMG), a phenotypic drug discovery-oriented approach that integrates cellular morphological profiles from compound treatments with molecular textual descriptions without requiring molecular target information. Cell morphology offers the guidance on desired bioactivity-relevant phenotypic effects, while textual descriptions provide direct and interpretable cues about molecular structure. Leveraging complementary structural and bioactivity context, MGMG significantly enhances molecule generation performance, especially in scenarios where textual descriptions are under-informative or morphological signals are weak. MGMG can also be applied to genetic perturbations, enabling activator design from gene overexpression morphology without requiring knowledge of reference compound structure. In addition, in silico docking demonstrates that MGMG-generated molecules, despite lacking target information, exhibit binding affinities comparable to reference compounds, preserving key interactions while introducing structural diversity. Overall, MGMG jointly utilizes morphological and textual description inputs to guide molecule generation, enabling diverse, bioactivity-aware compound design in a target-agnostic fashion.

  bioinformatics2026-08-01v2

• Pangenome Graph Node-Phenotype Association shows GWAS-like quality results with only few individuals

  Carrette, C.; Sabot, F.; Muller, C.

  Abstract

  Purpose: We introduce GraNPA, standing for Graph Node-Phenotype Associa- tion, a method performing a GWAS-like analysis on a pangenome variation graph (PVG) built using a small number of individual genome sequences, without the need for additional population materials or kinship information for qualitative phenotypes. This method reduces the number of individuals required for associ- ation studies and prevents reference bias from variant calling in these types of analyses. Background: A PVG represents the multiple alignment of a set of complete genomes. It contains all variations, from single nucleotide polymorphisms (SNPs) to large structural variations (SVs), which are represented as nodes in the graph. By integrating phenotype information within nodes, we can assign a Phenotype Score (PS) to each node in the PVG and identify phenotype-related regions directly within it. These regions represent statistically significant shifts in PS distribution, highlighting their implication in the phenotype. Finally, GraNPA provides their positions and scores for further analysis. Results: This method was tested using simulated data and two publicly available datasets: the Sub1A gene locus for Oryza sativa in a 13 indi- viduals PVG, and the insertion responsible for the white-headed cattle with a PVG of 24 individuals. Source code of GraNPA is available here https://forge.ird.fr/diade/graphgwas/granpa under GNU GPLv3. Conclusion: GraNPA was able to identify the expected area in two simulated datasets and the responsible loci for these two known traits using only a few 1 dozen complete genomes in these PVGs. While currently limited to qualitative phenotypes, this method opens the way to more efficient ones relying on PVGs and few individuals.

  bioinformatics2026-08-01v1

• DPCGS: a computational framework for linking GWAS to single-cell transcriptomics in complex traits and diseases

  Liu, C.; Yuan, B.; Shen, B.; Li, J.; Zhu, R.; Yang, P.; Wu, B.; Xuan, Y.; Yang, S.; Yang, N.; Ma, L.; Liu, Q.; Dai, S.; Zhang, Y.

  Abstract

  Complex traits and diseases arise from the interplay between genetic variation and cellular heterogeneity, making it essential to understand how genetic risk manifests at the cellular level. However, connecting genome-wide association studies (GWAS) to specific cell populations remains challenging due to cellular complexity and the prevalence of noncoding variants. Here, we present DPCGS, a computational framework that systematically integrates GWAS summary statistics with single-cell RNA-sequencing (scRNA-seq) data to identify trait-associated cell subpopulations, genes, and regulatory programs. Unlike existing approaches that primarily evaluate pathway enrichment or cell-type-level associations, DPCGS quantifies the enrichment of genetically prioritized genes within individual cells through a statistically calibrated gene-set scoring strategy, enabling high-resolution mapping of genetic risk to cellular states. Benchmarking across simulated and diverse human single-cell datasets demonstrates that DPCGS achieves superior accuracy, sensitivity, and robustness compared with existing methods, including scDRS and scPagwas. Applying DPCGS to Alzheimer's disease and asthma reveals disease-associated cellular populations and uncovers potential molecular drivers, including CD74, FOS, and AP-1 family regulatory programs, providing insights into disease-specific immune and cellular mechanisms. By bridging genetic discoveries from GWAS with functional interpretation at single-cell resolution, DPCGS establishes a generalizable framework for dissecting the cellular architecture of complex traits and diseases. This approach enables systematic discovery of disease-relevant cell subpopulations, regulatory networks, and potential therapeutic targets, offering broad applications in human genetics, single-cell biology, and precision medicine.

  bioinformatics2026-07-31v2