Bioinformatics (Oxford, England)[JOURNAL]
Sun
200 papers
microntology: a lightweight, data-driven controlled vocabulary to describe earth's microbial habitats.
Fullam A, Prasoodanan PKV, Kuhn M … +2 more , Bork P, Schmidt TSB
Bioinformatics · 2026 Jun · PMID 42203690 · Full text
MOTIVATION: Data-enabled studies of microbial ecology and evolution depend on high-quality descriptions of microbial habitats, based on curated and consolidated vocabularies. RESULTS: We introduce microntology v1.0, a pr...
MOTIVATION: Data-enabled studies of microbial ecology and evolution depend on high-quality descriptions of microbial habitats, based on curated and consolidated vocabularies. RESULTS: We introduce microntology v1.0, a pragmatic controlled vocabulary of 148 terms to describe microbial habitats and lifestyles, and provide manually curated microntology annotations for >300k metagenomic samples from public repositories. AVAILABILITY: microntology controlled vocabulary terms and term hierarchies (doi: 10.5281/zenodo.19730167), and curated annotations for 305 626 metagenomic samples (doi: 10.5281/zenodo.18164252) are available via Zenodo and spire.embl.de/downloads. Underlying code is available via github.com/grp-schmidt/microntology and Zenodo (doi: 10.5281/zenodo.20323497). User feedback, suggestions and bug reports are welcome at github.com/grp-schmidt/microntology/issues.
Lift&Add-rapid and robust addition of new species to alignments of conserved non-coding sequences.
Shukla N, Gallego Romero I
Bioinformatics · 2026 Jun · PMID 42203687 · Full text
MOTIVATION: Identifying sequence constraint across long evolutionary distances is a powerful method for the discovery of functional genomic sequences, especially putative non-coding elements. Conserved elements have been...
MOTIVATION: Identifying sequence constraint across long evolutionary distances is a powerful method for the discovery of functional genomic sequences, especially putative non-coding elements. Conserved elements have been a mainstay of comparative genomic research, and can be further investigated for species-specific sequence acceleration to dissect the genetic basis of trait evolution. The conclusions of these comparative genomic studies are contingent on the number and range of species included in this phylogenetic analysis. However, while the number of metazoan genomes sequences is increasing rapidly, adding new genomes to existing whole-genome alignments remains computationally expensive. RESULTS: Here, we present a bioinformatic workflow, Lift&Add, that enables conserved elements, coding or non-coding, to be rapidly mapped to new genomes ("Lift") and subsequently be added to pre-existing multiple species alignments ("Add"), thus providing an avenue for easy exploration of these putative functional elements. Focusing here on a group of species that has been largely under-represented in genomic comparisons, the marsupials, we demonstrate the intuition behind this workflow and provide an example comparative genomic analysis that can be performed. IMPLEMENTATION AND AVAILABILITY: Lift&Add is implemented as a series of scripts in Snakemake and bash, which can be downloaded from https://github.com/navyashukladr/Lift_and_Add.
Using cancer profiles to identify synthetic lethal therapeutic targets and predictive biomarkers in cancer gene dependency data.
Pearl LH, Pearl FMG
Bioinformatics · 2026 Jun · PMID 42203685 · Full text
MOTIVATION: Large scale loss-of-function screens utilising CRISPR or siRNA can provide profound insights into the importance of individual genes for the survival of a cancer cell and can drive the identification of thera...
MOTIVATION: Large scale loss-of-function screens utilising CRISPR or siRNA can provide profound insights into the importance of individual genes for the survival of a cancer cell and can drive the identification of therapeutic targets and biomarkers, and the development of targeted drugs. However, the analysis of these data and the substantial bodies of metadata that relate to them, is technically challenging and typically requires substantial expertise in data science and computer coding. RESULTS: To facilitate the analysis of cancer gene dependency data by cancer biologists and clinical scientists, we have developed DepMine-a computational toolkit providing a powerful system for framing complex queries relating cancer gene dependency to the underlying genetic changes that occur in cancer cells. DepMine identifies synthetic lethal relationships between putative target genes and complex 'cancer profiles' built from user-specified combinations of mutations, copy-number variation, and expression levels, and can refine these to optimal biomarker definitions for target dependency. AVAILABILITY: The Python implementation of DepMine and associated data files can be obtained at https://github.com/UOSbioinformaticslab/depmine and is free to academics and Not-For-Profit organisations. The DepMine release referenced in this paper is archived as DOI: 10.5281/zenodo.19570601.
mmContext: an open framework for multimodal contrastive learning of omics and text data.
Menger J, Krissmer SM, Kreutz C … +2 more , Binder H, Hackenberg M
Bioinformatics · 2026 Jun · PMID 42191659 · Full text
SUMMARY: Multimodal approaches are increasingly leveraged for integrating omics data with textual biological knowledge. Yet there is still no accessible, standardized framework that enables systematic comparison of omics...
SUMMARY: Multimodal approaches are increasingly leveraged for integrating omics data with textual biological knowledge. Yet there is still no accessible, standardized framework that enables systematic comparison of omics representations with different text encoders within a unified workflow. We present mmContext, a lightweight and extensible multimodal embedding framework built on top of the open-source Sentence Transformers library. The software allows researchers to train or apply models that jointly embed omics and text data using any numeric representation stored in an AnnData.obsm layer and any text encoder available in Hugging Face. mmContext supports integration of diverse biological text sources and provides pipelines for training, evaluation, and data preparation. We train and evaluate models for a RNA-Seq and text integration task, and demonstrate their utility through zero-shot classification of cell types and diseases across four independent datasets. By releasing all models, datasets, and tutorials openly, mmContext enables reproducible and accessible multimodal learning for omics-text integration. AVAILABILITY AND IMPLEMENTATION: Pretrained checkpoints and full source code for our custom MMContextEncoder are available on Hugging Face huggingface.co/jo-mengr. The Python package github.com/mengerj/mmcontext provides the model implementation and training and evaluation scripts for custom training. The releases for the publication can be accessed via zenodo: adata_hf_datasets: doi.org/10.5281/zenodo.19185217 and mmContext: doi.org/10.5281/zenodo.19185493.
A variational framework with composite sparse regularization for cryo-electron tomography reconstruction.
Yu C, Xu Z, Zeng Q … +4 more , Wan X, El-Messiry H, Zhang F, Han R
Bioinformatics · 2026 Jun · PMID 42191651 · Full text
MOTIVATION: Cryogenic electron tomography (cryo-ET) enables in situ visualization of macromolecular and cellular structures from tilt-series projections. Reconstruction quality is often compromised by extremely low signa...
MOTIVATION: Cryogenic electron tomography (cryo-ET) enables in situ visualization of macromolecular and cellular structures from tilt-series projections. Reconstruction quality is often compromised by extremely low signal-to-noise ratio (SNR) and vignetting artifacts arising from detector truncation under constrained acquisition geometries. In practice, existing methods frequently struggle to balance noise robustness, computational efficiency, and stability under these conditions. RESULTS: We propose a robust, scalable, and parallelizable variational reconstruction framework that integrates a geometrically consistent data fidelity term with an implicit boundary-handling mechanism to mitigate truncation-induced artifacts without volume padding. A composite sparse regularizer integrating anisotropic total variation and curvelet-domain sparsity is employed to preserve structural boundaries and multiscale directional features. The resulting optimization problem is efficiently solved using the primal-dual hybrid gradient (PDHG) algorithm without nested inner iterations, for which we provide rigorous theoretical guarantees of stability and convergence. Experiments on simulated and experimental cryo-ET datasets demonstrate substantial noise suppression and contrast enhancement while preserving fine structural details under realistic, severely noise-limited and truncated acquisition conditions. These improvements lead to enhanced interpretability and facilitate downstream structural analysis, while achieving significantly reduced runtime compared to existing methods at comparable reconstruction quality. AVAILABILITY AND IMPLEMENTATION: Our code available at https://github.com/icthrm/CSRT. The real datasets used in this study are publicly available from EMPIAR and the Caltech Electron Tomography Database.
MolPIF: a parameter interpolation flow model for molecule generation.
Jin Y, Wang J, Tang Y … +11 more , Xiang W, Cao D, Teng D, Fan Z, Xiong J, Sheng X, Zeng C, An D, Zheng M, Zheng S, Shi Q
Bioinformatics · 2026 Jun · PMID 42184281 · Full text
MOTIVATION: Structure-based drug design (SBDD) has advanced with deep generative models, but bridging the gap between continuous atomic coordinates and discrete atom types remains a challenge. Current approaches, such as...
MOTIVATION: Structure-based drug design (SBDD) has advanced with deep generative models, but bridging the gap between continuous atomic coordinates and discrete atom types remains a challenge. Current approaches, such as diffusion and flow matching models, often fail to unify these heterogeneous modalities, relying on separate strategies or ill-fitting Euclidean metrics for discrete variables. This lack of a consistent framework limits generative models' ability to capture the geometric and chemical structure of protein-ligand complexes. RESULTS: We present MolPIF, a parameter interpolation flow mechanism designed to unify the generation of continuous and discrete molecular variables. Unlike traditional flow models that operate in sample space, MolPIF interpolates between distributions in the parameter space, theoretically recovering Wasserstein-2 optimal transport for continuous coordinates and establishing Fisher-Rao geodesics for discrete atom types. We further incorporate a geometry-enhanced learning strategy to improve the capture of atomic contexts. Extensive evaluations on the CrossDocked2020 dataset demonstrate that MolPIF outperforms baselines in binding affinity, chemical validity, geometric fidelity, and chemical space coverage. Additionally, MolPIF exhibits versatility in lead optimization and offers flexible prior distribution selection (such as Laplace), establishing a robust paradigm for SBDD. AVAILABILITY AND IMPLEMENTATION: Source code is freely available at https://github.com/BLEACH366/MolPIF.
HKD-CPI: high-order knowledge distillation enhanced inductive compound-protein interaction prediction.
He Z, Liu X, Jiang Y … +5 more , Xu J, Lin Y, Jin S, Wei L, Wang Y
Bioinformatics · 2026 Jun · PMID 42179166 · Full text
MOTIVATION: Accurately identifying compound-protein interactions (CPIs) is critical for accelerating drug discovery. Recent deep learning methods have achieved impressive results, yet they primarily focus on local struct...
MOTIVATION: Accurately identifying compound-protein interactions (CPIs) is critical for accelerating drug discovery. Recent deep learning methods have achieved impressive results, yet they primarily focus on local structures and neighborhood information, often overlooking high-order interaction patterns shared among similar molecules. RESULTS: In this paper, we propose HKD-CPI, a high-order knowledge-enhanced inductive framework designed to improve generalization to unseen compound-protein pairs. Specifically, HKD-CPI introduces a molecular graph tokenization mechanism that aligns compound molecular graph features with token embeddings from sequence-pretrained large language models (LLMs), effectively infusing sequence-derived semantics into structural representations. To capture shared interaction patterns among functionally similar biomolecules, we construct a hypergraph-based representation to model high-order relationships between feature-similar compound/protein groups and their binding partners. Furthermore, a knowledge distillation strategy is further adopted to transfer high-order interaction knowledge from the hypergraph to a lightweight student model, enabling efficient and robust CPI prediction. Extensive experiments demonstrate that HKD-CPI outperforms existing state-of-the-art methods in inductive CPI prediction tasks. In particular, it achieves an average improvement of 4.94% in AUROC and 3.64% in AUPRC over the best-performing baseline across five benchmark datasets. AVAILABILITY AND IMPLEMENTATION: Our code and data are available at https://github.com/Hezy618/HKD-CPI.
slideimp: efficient imputation of DNA methylation data.
Pham H, Lombroso AP, Cevik EC … +2 more , Taylor HS, O'Donnell KJ
Bioinformatics · 2026 Jun · PMID 42179165 · Full text
SUMMARY: We developed slideimp, an R package that extends and optimizes K-nearest neighbor (K-NN) and Principal Component Analysis (PCA) imputation with grouped and sliding-window modes for accurate and efficient imputat...
SUMMARY: We developed slideimp, an R package that extends and optimizes K-nearest neighbor (K-NN) and Principal Component Analysis (PCA) imputation with grouped and sliding-window modes for accurate and efficient imputation of microarray and whole-genome DNA methylation (DNAm) data, respectively. Under a realistic scenario, slideimp achieved ≈12-28× faster runtime and ≈3-6× peak memory usage reduction for DNAm microarray imputation (GSE286313, EPICv2, N = 72) and achieved high imputation accuracy in a whole-genome DNAm dataset (N = 41). AVAILABILITY AND IMPLEMENTATION: The code used in this study is available at https://github.com/hhp94/slideimp_paper. The R package slideimp is available on CRAN (DOI: 10.32614/CRAN.package.slideimp). Version 1.0.0 of slideimp, which was used in this study, is archived on Zenodo (DOI: 10.5281/zenodo.20029382).
Representation learning for multi-modal spatially resolved transcriptomics data.
Nonchev K, Andani S, Ficek-Pascual J … +5 more , Nowak M, Sobottka B, Consortium TP, Koelzer VH, Rätsch G
Bioinformatics · 2026 May · PMID 42179160 · Publisher ↗
MOTIVATION: Spatial transcriptomics enables in-depth molecular characterization of samples on a morphology and RNA level while preserving spatial location. Integrating the resulting multi-modal data is an unsolved proble...
MOTIVATION: Spatial transcriptomics enables in-depth molecular characterization of samples on a morphology and RNA level while preserving spatial location. Integrating the resulting multi-modal data is an unsolved problem, and developing new solutions in precision medicine depends on improved methodologies. RESULTS: We introduce AESTETIK, a convolutional deep learning model that jointly integrates spatial, transcriptomics, and morphology information to learn accurate spot representations. AESTETIK yielded substantially improved cluster assignments on widely adopted technology platforms (e.g. 10x Genomics™, NanoString™) across multiple datasets. We achieved performance enhancement on structured tissues (e.g. brain) with a 21% increase in median ARI over previous state-of-the-art methods. Notably, AESTETIK also demonstrated superior performance on cancer tissues with heterogeneous cell populations, showing a two-fold increase in breast cancer, 79% in melanoma, and 21% in liver cancer. We expect that these advances will enable a multi-modal understanding of key biological processes. AVAILABILITY AND IMPLEMENTATION: AESTETIK is implemented in Python 3 and is available as open source software at http://www.github.com/ratschlab/aestetik. The Snakemake pipeline for reproducing the results is available at http://www.github.com/ratschlab/st-rep. CONTACT: kalin.nonchev@inf.ethz.ch, viktor.koelzer@usb.ch, gunnar.raetsch@inf.ethz.ch. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
MetaStrainer: accurate reconstruction of bacterial strain genotypes from short-read metagenomic samples.
Sharaf H, Bobay LM
Bioinformatics · 2026 Jun · PMID 42178395 · Full text
MOTIVATION: Metagenomics provides broad insights from microbial communities, but more biological relevant phenotypes are attributed to subtle changes at the strain-level rather than species. Despite development of severa...
MOTIVATION: Metagenomics provides broad insights from microbial communities, but more biological relevant phenotypes are attributed to subtle changes at the strain-level rather than species. Despite development of several tools using different algorithms, resolving individual strains from short-read pair-end sequencing data remains challenging. RESULTS: Here we present MetaStrainer, a tool capable of reconstructing strain genotypes from metagenomic data. Compared with existing approaches, MetaStrainer substantially increases genotype accuracy, correctly identifies the number of strains, and accurately estimates their relative abundances. Accuracy of reconstructed genotypes is robust to choice of mapping reference. AVAILABILITY: MetaStrainer is implemented in Python 3. Source code and instructions are available on GitHub at www.github.com/lbobay/MetaStrainer and on Zenodo: 10.5281/zenodo.17872331.
Sassy: fuzzy searching DNA sequences using SIMD.
Beeloo R, Groot Koerkamp R
Bioinformatics · 2026 May · PMID 42178392 · Full text
MOTIVATION: Approximate string matching (ASM) is the problem of finding all occurrences of a pattern in a text while allowing up to k errors. Many modern methods use seed-chain-extend, which is fast in practice, but does...
MOTIVATION: Approximate string matching (ASM) is the problem of finding all occurrences of a pattern in a text while allowing up to k errors. Many modern methods use seed-chain-extend, which is fast in practice, but does not guarantee finding all matches with ≤k errors. However, applications such as CRISPR off-target detection require exhaustive results. RESULTS: We introduce Sassy, a library and tool for ASM of short patterns in long texts. Sassy splits the text into four parts that are searched in parallel, and uses bitvectors in the text direction rather than the pattern direction. This has complexity O(k⌈n/W⌉) when searching a random text of length n, where W=256 is the SIMD width, and provides significant speedups for small k. Separately, we allow matches of the pattern to extend beyond the text for an overhang cost of, e.g. α=0.5 per character, to find matches near contig or read ends.Sassy is 4× to 15× faster than Edlib for patterns ≤1000 bp, and can search text with a throughput near 2 Gbp/s. Likewise, Sassy is over 100× faster than parasail. We apply Sassy to CRISPR off-target detection by searching 61 guide sequences in a human genome. Sassy is 100× faster than SWOffinder and only slightly slower (for k≤3) than CHOPOFF, for which building its index takes 20 min. Sassy also scales well to larger k, unlike CHOPOFF whose index took over 10 h to build for k=5. AVAILABILITY AND IMPLEMENTATION: Sassy is available as library and binary at https://github.com/RagnarGrootKoerkamp/sassy, and archived at swh:1:dir:e884758dce5777a441bc2799dc8824e563c5f97b.
Control-guided refinement of partially specified Boolean networks: applications to RTK signaling.
Šmijáková E, Brim L, Pastva S … +1 more , Šafránek D
Bioinformatics · 2026 Jun · PMID 42178383 · Full text
MOTIVATION: System control can be used to provide new insights into the dynamics of biological systems. A key application is the identification of therapeutic targets in silico, which requires an executable model of the...
MOTIVATION: System control can be used to provide new insights into the dynamics of biological systems. A key application is the identification of therapeutic targets in silico, which requires an executable model of the system's dynamics. However, such models are typically underspecified due to incomplete mechanistic knowledge. RESULTS: We introduce a novel computational framework that employs control-guided model refinement, predicting informative perturbation experiments to reduce knowledge gaps. The approach is based on partially specified Boolean networks (PSBNs), which enable direct integration of uncertain or incomplete information into executable models. We further extend the framework to handle oscillatory phenotypes as explicit control targets. The applicability of the method is demonstrated on receptor-tyrosine kinase (RTK) signaling, with a focus on fibroblast growth factor signaling in the context of skeletal dysplasias and cancer. We obtain several new insights into modelling of the FGFR3-MAPK pathway. AVAILABILITY AND IMPLEMENTATION: Code and datasets are available at https://doi.org/10.5281/zenodo.16886813.
MarkerMatch: a proximity-based probe-matching algorithm for joint analysis of copy-number variants from different genotyping arrays.
Ivankovic F, Yu D, Shen J … +10 more , Zhan L, Niarchou M, Kaylor A, Domènech L, Miller-Fleming TW, Porras LM, Giusti-Rodríguez P, Ophoff RA, Scharf JM, Mathews CA
Bioinformatics · 2026 Jun · PMID 42178371 · Full text
MOTIVATION: Copy-number variants (CNVs) are a form of genetic structural variation with increasing importance in complex human disorders. Both DNA sequencing and microarray data can be used to detect CNVs, which can be u...
MOTIVATION: Copy-number variants (CNVs) are a form of genetic structural variation with increasing importance in complex human disorders. Both DNA sequencing and microarray data can be used to detect CNVs, which can be used in genetic association tests. Unlike genotypes, CNV detection in microarrays requires the use of observed intensity signals at each probe, which limits the imputability for analyses that span multiple array types. Thus far, a consensus set of probes (those present on all arrays) has been used to circumvent the problem of differing array-specific sensitivities. This has led to excessive reduction in overall sensitivity since arrays can have an undesirably low probe overlap. To overcome this limitation, we developed MarkerMatch, a proximity-based algorithm that matches probes across different genotyping microarrays to maximize the number of probes considered in the CNV calling algorithm, thereby increasing the resolution and sensitivity while preserving precision. RESULTS: By analyzing CNV calls from 4906 individuals genotyped across three different arrays, we show that the MarkerMatch approach improves sensitivity by increasing the density of probes available for CNV calling while maintaining precision or improving it relative to the current practice (e.g. use of consensus probes only). We further demonstrate that MarkerMatch matches the CNV detection from current practice in terms of F1 score and PPV for larger CNVs. We also optimize MarkerMatch parameters, DMAX and Method, and find an optimal DMAX setting at 10 kb, with no clear optimal candidate based on Method, indicating that parameters for this metric should be determined on a use case basis. AVAILABILITY: The R package for MarkerMatch is available at: https://github.com/FranjoIM/MarkerMatch. The code used for analysis and implementation is available at: https://doi.org/10.5281/zenodo.18460979. The live notebook is available at https://fivankovic.notion.site/2026-markermatch.
ChromCall: assigning chromatin status to defined genomic regions using epigenomic profiling data.
Wang B, Al-Jabri M, Gk U … +2 more , Droop A, Stead LF
Bioinformatics · 2026 Jun · PMID 42178345 · Full text
MOTIVATION: Chromatin regulation is crucial for modulating gene expression and cellular function by altering DNA accessibility. Defining and understanding chromatin regulation across diverse biological conditions, includ...
MOTIVATION: Chromatin regulation is crucial for modulating gene expression and cellular function by altering DNA accessibility. Defining and understanding chromatin regulation across diverse biological conditions, including health and disease, requires quantification of both the presence and enrichment level of diverse DNA-binding factors and chromatin modifications across defined genomic regions. Existing approaches mainly rely on peak-based or genome-wide models, which identify high-signal regions but do not annotate chromatin status at predefined functional genomic regions, such as promoters or enhancers. This lack of region-based annotation limits downstream comparative and integrative analyses across multiple factors and datasets, prompting us to create ChromCall. RESULTS: ChromCall is an R package for region-based chromatin enrichment analysis that provides a robust and extensible foundation for transparent and reproducible epigenomic profiling at predefined genomic regions. We applied ChromCall to ChIP-seq data from glioblastoma (GBM) brain tumours and found that the promoters of genes implicated in treatment resistance are significantly more likely to exhibit a combination of histone marks associated with phenotypic plasticity. This highlights a potential novel mechanism of therapeutic escape in these deadly tumours. AVAILABILITY AND IMPLEMENTATION: The R package is available on https://github.com/GliomaGenomics/ChromCall and the version used in this paper is archived at https://doi.org/10.5281/zenodo.19580967.
A masked generative graph representation learning framework empowering precise spatial domain identification.
Wang C, Zhang T, Sun H … +11 more , Wu Z, Liang S, Wang X, Du M, Liang Y, Gao X, Tang Q, Xu D, Feng X, Zeng A, Guan R
Bioinformatics · 2026 Jun · PMID 42178226 · Full text
MOTIVATION: Spatial transcriptomics (ST) enables the measurement of gene expression while preserving the spatial context of tissues. However, the sparsity of ST data leads to poor usage of gene expression and spatial inf...
MOTIVATION: Spatial transcriptomics (ST) enables the measurement of gene expression while preserving the spatial context of tissues. However, the sparsity of ST data leads to poor usage of gene expression and spatial information, resulting in the embeddings that are not well represented and challenging for downstream analyses. RESULTS: Here, we introduced GSG, a generative self-supervised representation learning framework for ST data that leverages a masking mechanism to learn informative representations. For spatial domain identification, GSG consistently outperformed state-of-the-art methods across benchmarking datasets, regardless of sequencing platforms. In addition, we applied GSG to an in-house human fetal heart dataset, revealing anatomically coherent spatial domains and identifying APCDD1 as an endocardial-specific marker potentially involved in congenital heart disease. Our results showcase GSG's superiority and underscore its valuable contributions to advancing ST analysis. AVAILABILITY AND IMPLEMENTATION: Our software package is available at https://github.com/keaml-Guan/GSG.
Trimmomatic: a decade of feature-rich, high-performance NGS read preprocessing.
Beier S, Bolger AM, Bolger ME … +2 more , Schwacke R, Usadel B
Bioinformatics · 2026 Jun · PMID 42178219 · Full text
MOTIVATION: Trimmomatic is a widely adopted tool for preprocessing high-throughput sequencing data, particularly from Illumina platforms. Since its original publication in 2014, the volume and complexity of sequencing da...
MOTIVATION: Trimmomatic is a widely adopted tool for preprocessing high-throughput sequencing data, particularly from Illumina platforms. Since its original publication in 2014, the volume and complexity of sequencing data have increased dramatically, necessitating continuous tool evolution. RESULTS: We present the substantial updates to Trimmomatic over the past decade. Key enhancements include a robust multithreading model for high-performance parallel processing, parallel GZIP/BZIP2 compression, and a suite of new trimming and filtering steps to provide users with more flexible quality control. Usability has been significantly improved through automatic PHRED encoding detection and simplified file handling. The codebase has also been modernized including Maven support, and continuous integration to ensure long-term sustainability and community contributions. These updates solidify Trimmomatic's role as an efficient, flexible, and essential tool in modern bioinformatics pipelines. AVAILABILITY: Trimmomatic remains open-source under the GPL V3 license, with the latest version available at https://github.com/usadellab/Trimmomatic and also on our website https://www.plabipd.de/trimmomatic_main.html (DOI: https://doi.org/10.5281/zenodo.18678155).
DA-BioNER: data augmentation based on few-shot learning and distant supervision for biomedical named entity recognition.
Park Y, Son G, Kim T … +1 more , Rho M
Bioinformatics · 2026 Jun · PMID 42178206 · Full text
MOTIVATION: Named entity recognition (NER) is a fundamental component of structured knowledge extraction, yet its effectiveness in emerging domains remains by the scarcity of high-quality, domain-specific annotated corpo...
MOTIVATION: Named entity recognition (NER) is a fundamental component of structured knowledge extraction, yet its effectiveness in emerging domains remains by the scarcity of high-quality, domain-specific annotated corpora. Although data augmentation and distant supervision have been explored to alleviate this issue, existing methods often introduce limited entity diversity, noisy labels, or disrupt contextual integrity, thereby limiting their generalization ability in low-resource settings. RESULTS: In this study, we propose DA-BioNER, a context-preserving data expansion framework for biomedical NER. DA-BioNER combines multiple base NER models trained on few-shot data to provide coarse annotations, followed by refinement using a large language model (LLM) guided by global biomedical knowledge. Unlike generation-based augmentation methods that synthesize new sentences, DA-BioNER performs annotation refinement within existing sentences, preserving both syntactic structure and semantic context. By constraining the role of LLM to refinement rather than open-ended generation, the framework effectively reduces hallucination while improving label precision and consistency. We evaluate DA-BioNER on three benchmark datasets (NCBI-Disease, BC5CDR, and BioRED), under low-resource conditions. In 40-shot settings, DA-BioNER achieves F1-scores of 0.750, 0.795, and 0.799, respectively, outperforming state-of-the-art methods, including LSMS, DAGA, and MELM, by up to 0.32. Under more extreme few-shot settings, DA-BioNER further improves F1-scores by up to 0.08, while generating an average of 1,391 additional unique entities, substantially enriching training diversity. These results demonstrate that DA-BioNER provides a scalable and adaptable solution for robust biomedical NER, particularly in domain adaptation and low-resource scenarios. AVAILABILITY: DA-BioNER is publicly available at https://github.com/DMnBI/DA-BioNER.
Clarifying the scope and capabilities of ROTS in differential expression analysis.
Suomi T, Kettunen J, Pusa T … +1 more , Elo LL
Bioinformatics · 2026 Jun · PMID 42178203 · Full text
SUMMARY: Recently, Anwar et al. introduced a method combining the ROTS reproducibility optimisation procedure with empirical Bayes variance estimation from limma. Here, we clarify several methodological aspects to suppor...
SUMMARY: Recently, Anwar et al. introduced a method combining the ROTS reproducibility optimisation procedure with empirical Bayes variance estimation from limma. Here, we clarify several methodological aspects to support accurate interpretation of the results. We emphasise that ROTS is a general reproducibility optimisation framework rather than a single statistical test and demonstrate that benchmarking outcomes in the reported spike-in case studies are highly sensitive to analysis and evaluation choices. Furthermore, our reanalyses of the spike-in datasets do not support the reported conclusions, and we were unable to reproduce the results of the clinical Alzheimer's disease case study. These findings highlight the importance of transparent benchmarking practices and careful interpretation of comparative results. AVAILABILITY AND IMPLEMENTATION: The ROTS package is available through Bioconductor. The reanalyses were performed using the original code, with the minimal additions described in the manuscript.
Atlas-level single-cell integration and clustering-free differential expression analysis with GEDI 2.0.
Mikaeili Namini A, Saberi A, Najafabadi HS
Bioinformatics · 2026 Jun · PMID 42178201 · Full text
MOTIVATION: GEDI is a generative framework for multi-sample, multi-condition single-cell analysis that performs batch correction, latent representation learning, and clustering-free differential expression within a unifi...
MOTIVATION: GEDI is a generative framework for multi-sample, multi-condition single-cell analysis that performs batch correction, latent representation learning, and clustering-free differential expression within a unified model. However, the original implementation suffered from prohibitive memory use and runtime, preventing its application to modern atlas-scale datasets. RESULTS: We present GEDI 2.0, a complete high-performance reimplementation featuring a standalone C++ computational core with pre-allocated workspaces, strict sparse-matrix preservation, optimized BLAS routines, and multi-threaded block-coordinate descent. Across extensive benchmarks spanning up to 500 000 cells and 10 000 features, GEDI 2.0 achieves 40%-63.6% mean reduction in peak memory, 2.98× mean single-threaded speedups, and up to 11.5× acceleration with parallel execution, while maintaining full numerical equivalence to the original method. These improvements enable GEDI 2.0 to analyze million-cell datasets, a scale not achievable with the legacy implementation. GEDI 2.0 provides R and Python interfaces and seamless interoperability with common single-cell workflows. AVAILABILITY AND IMPLEMENTATION: Source code, documentation, reproducible codebase, and tutorials are available at https://github.com/csglab/gedi2.
GLiNER-BioMed: a suite of efficient models for open biomedical named entity recognition.
Yazdani A, Stepanov I, Teodoro D
Bioinformatics · 2026 Jun · PMID 42172599 · Full text
MOTIVATION: Biomedical named entity recognition (NER) presents unique challenges due to specialized vocabularies, the sheer volume of entities, and the continuous emergence of novel entities. Traditional NER models, cons...
MOTIVATION: Biomedical named entity recognition (NER) presents unique challenges due to specialized vocabularies, the sheer volume of entities, and the continuous emergence of novel entities. Traditional NER models, constrained by fixed taxonomies and human annotations, struggle to generalize beyond predefined entity types. RESULTS: To address these issues, we introduce GLiNER-BioMed, a domain-adapted suite of GLiNER models for biomedicine. Our approach first distills the annotation capabilities of large language models (LLMs) into a smaller, more efficient model, enabling the generation of high-coverage biomedical NER data. We subsequently train two GLiNER architectures, uni- and bi-encoder, at multiple scales to balance computational efficiency and performance. Experiments on eight biomedical datasets demonstrate that GLiNER-BioMed achieved state-of-the-art zero-shot performance (micro-F1 59.77%), exceeding the strongest baseline by 5.96 points (P < .001). In few-shot learning, the bi-encoder variant reached 70.39% (10-shot), consistently outperforming the strongest baseline across all settings (P < .05). Our findings show that the uni-encoder GLiNER-BioMed achieves the strongest zero-shot performance, while the bi-encoder offers superior few-shot gains and substantially higher inference throughput (+39%-568%), making it well-suited to annotation-limited, latency-sensitive, or large-label-space settings. Ablation studies further indicate that combining synthetic biomedical pre-training with general-domain post-training is essential for capturing domain-specific knowledge while maintaining precision-recall balance. AVAILABILITY AND IMPLEMENTATION: The source code, datasets, and models are publicly available at https://github.com/ds4dh/GLiNER-biomed.
About
- Frequency
- Sun
- Papers found
- 200
- RSS feed
- Subscribe