Bioinformatics (Oxford, England)[JOURNAL]
Sun
200 papers
Cross-Domain Transfer Learning from Peptides to Metabolites Using a Multi-Property Fine-Tuned LLM.
Anyaegbunam UA, Teschner D, Schmidlin T … +4 more , Hildebrandt A, Mayer JU, Sprang M, Andrade-Navarro MA
Bioinformatics · 2026 Jul · PMID 42402205 · Publisher ↗
MOTIVATION: Accurate liquid chromatography retention time (RT) prediction is a critical component of compound identification in metabolomics and lipidomics. However, existing RT prediction approaches are often limited by...
MOTIVATION: Accurate liquid chromatography retention time (RT) prediction is a critical component of compound identification in metabolomics and lipidomics. However, existing RT prediction approaches are often limited by the scarcity of experimental RT measurements for many molecular classes, restricting model generalization and the construction of comprehensive RT libraries. Transfer learning from data-rich chemical domains offers a potential strategy to overcome these limitations, but its effectiveness for metabolite RT prediction remains insufficiently explored. RESULTS: We developed a transfer learning framework based on ChemBERTa that leverages large peptide datasets to improve metabolite RT prediction under data-sparse conditions. A peptide-pretrained model was trained using a multi-task objective that jointly predicted RT and seven RDKit-derived molecular descriptors. Compared with an RT-only model, the multi-task approach learned more robust chemical representations and demonstrated superior generalization to metabolites, achieving a median test R² of 0.842 versus 0.820. When transferred to metabolite RT prediction, the multi-task pretrained model substantially outperformed models trained from scratch at low-data regimes. Using only 3% of metabolite training data (2,129 compounds), transfer learning achieved a median test R² of 0.322 compared with 0.216 for the baseline model, while reducing MAE from 131.7 to 114.9. Significant improvements were also observed at 5% and 10% training fractions, with benefits gradually diminishing as larger metabolite datasets became available. In contrast, a peptide-pretrained single-task RT model showed performance comparable to the baseline, indicating that the observed gains arise primarily from multi-task molecular property learning rather than peptide pretraining alone. These findings demonstrate that multi-task transfer learning provides an effective and scalable strategy for improving RT prediction in metabolomics, particularly when experimental training data are limited. AVAILABILITY: Freely available on https://github.com/uchealex/CHEMBEDDING. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
Biomedical Concept Recognition with Error-aware Negative-enhanced Ranking Framework.
Liu S, Nishida N, Cheng F … +2 more , Takehito U, Matsumoto Y
Bioinformatics · 2026 Jul · PMID 42398028 · Publisher ↗
MOTIVATION: Mention-agnostic biomedical concept recognition (MA-BCR) requires inferring ontology concepts directly from passages, without relying on explicit mention spans. Prior work has mainly focused on generative and...
MOTIVATION: Mention-agnostic biomedical concept recognition (MA-BCR) requires inferring ontology concepts directly from passages, without relying on explicit mention spans. Prior work has mainly focused on generative and classification-based approaches. Ranking-based methods typically use a retrieve-rerank pipeline, and this paradigm has not been systematically studied for MA-BCR. Consequently, it remains unclear how ranking-based approaches compare with existing paradigms and what types of supervision are most beneficial for ranker training under limited annotation settings. RESULTS: Through a systematic comparison of ranking-, generative-, and classification-based paradigms, we show that a two-stage retrieve-rerank architecture is the most robust and scalable backbone for MA-BCR. Building on this finding, we propose ENR, an error-aware negative-enhanced ranking framework that augments training with false positives collected from heterogeneous recognizers, improving reranking performance without increasing inference-time cost. Experiments on MM-HPO and MM-GO (two datasets derived from MedMentions-ST21pv) demonstrate that ENR substantially outperforms prior approaches. AVAILABILITY AND IMPLEMENTATION: Code and data are available at https://github.com/sl-633/enr-recognizer or https://doi.org/10.5281/zenodo.20730803. SUPPLEMENTARY INFORMATION: Supplementary materials are available at Supplemental_Materials.pdf.
TEDLH: Domain HMMs for sensitive detection of remote homologues.
Alvarez-Carreño C, Petrov AS, Waman VP … +2 more , Sillitoe I, Orengo C
Bioinformatics · 2026 Jul · PMID 42398027 · Publisher ↗
MOTIVATION: The Encyclopedia of Domains (TED) provides domain annotations for proteins in the AlphaFold Protein Structure Database (AFDB) using a consensus of three state-of-the-art structure-based methods. We used these...
MOTIVATION: The Encyclopedia of Domains (TED) provides domain annotations for proteins in the AlphaFold Protein Structure Database (AFDB) using a consensus of three state-of-the-art structure-based methods. We used these annotations to construct profile Hidden Markov models (HMMs), collectively forming the TED Library of HMMs (TEDLH). TEDLH enables sensitive sequence and profile searches, supporting systematic exploration of protein domain families and their evolutionary relationships. RESULTS: TEDLH links 934,186 domain HMMs to experimentally determined CATH-PDB structures through direct (primary) and transitive (secondary and tertiary) relationships. Fewer than half of TEDLH HMMs are directly linked to a CATH-PDB domain; the remaining models are connected through transitive relationships. These transitive links extend coverage into more divergent regions of sequence space and better represent CATH superfamily diversity.HMM-HMM comparisons within CATH superfamily 3.30.70.100 illustrate how transitive relationships expand sequence coverage. In this superfamily, 5,640 TEDLH HMMs are connected to 173 CATH-PDB representatives. Primary, secondary, and tertiary relationships progressively capture more divergent sequences (pairwise sequence identity <20%) that retain structural similarity (TM-score ≥0.6) and a conserved two-layer α/β sandwich core fold.All-against-all HMM-HMM comparisons across TEDLH also reveal sequence similarities across the CATH hierarchy (cross-hits). At low query coverage (<50%), cross-hits are more frequent between CATH classes, architectures and topologies, whereas at higher coverage thresholds (≥70%) they predominantly occur between superfamilies. These cross-hits are not driven by superfamily size or sequence diversity and can provide guidance for CATH curation. As an example, analysis of cross-hits between superfamilies 2.170.130.30 and 3.10.20.30 reveals evolutionary relationships between these groups. AVAILABILITY: TEDLH is compatible with HH-suite3 and is available from FigShare https://doi.org/10.6084/m9.figshare.28531754 for local use. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
PLNFGL: Joint Estimation of Multi-Condition Gene Networks from Single-cell RNA-seq Data.
Zhai W, Zhou D, Yuan Z … +1 more , Ji J
Bioinformatics · 2026 Jul · PMID 42398025 · Publisher ↗
MOTIVATION: Graphical models have been widely used in bioinformatics to infer the conditional dependence structure among random variables, but traditional Gaussian graphical models (GGMs) are suboptimal for single-cell R...
MOTIVATION: Graphical models have been widely used in bioinformatics to infer the conditional dependence structure among random variables, but traditional Gaussian graphical models (GGMs) are suboptimal for single-cell RNA sequencing (scRNA-seq) due to dropout events and distributional mismatch. Moreover, most existing methods estimate networks under a single condition, limiting their utility in multi-condition studies. RESULTS: We propose PLNFGL (Poisson Log-Normal Fused Graphical Lasso), a joint network estimation framework for scRNA-seq data. PLNFGL uses a multivariate Poisson log-normal model to accommodate dropout effects and estimates the covariance via moment methods. A joint graphical model is then employed to infer condition-specific precision matrices. Simulations show improved estimation accuracy. Applications to scRNA-seq data of Alzheimer's disease and spatial transcriptomics of lung cancer reveal cell-type-specific interaction networks. Edge set enrichment enables pathway analysis, validating known interactions and highlighting novel disease-related targets. This work provides a powerful tool for the integrative analysis of scRNA-seq data. AVAILABILITY AND IMPLEMENTATION: The R implementation of PLNFGL is available at https://github.com/jijiadong/PLNFGL, and an archival version is available on Zenodo at https://doi.org/10.5281/zenodo.20744172. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
MCFST: Spatial domain identification method based on multi-view graph convolutional network and graph fusion network.
Zhang Z, Duan H, Gao X
Bioinformatics · 2026 Jul · PMID 42397221 · Publisher ↗
MOTIVATION: The emergence of spatial transcriptomics, which integrates spatial and gene expression information, has greatly advanced research in disease mechanisms and developmental biology. A core task in this field is...
MOTIVATION: The emergence of spatial transcriptomics, which integrates spatial and gene expression information, has greatly advanced research in disease mechanisms and developmental biology. A core task in this field is spatial domain identification, which reveals regions with shared molecular signatures and histological features, thereby facilitating the study of tissue function and pathology. Although existing methods have achieved promising performance, many of them still face limitations in effectively integrating heterogeneous information from multiple views, such as gene expression, spatial coordinates, and spatially informed expression profiles. In particular, discrepancies across views may lead to inconsistent representations and distorted similarity relationships, which can reduce the accuracy and robustness of spatial domain recognition. RESULTS: To address these limitations, we propose MCFST, a graph neural network framework that integrates multi-view graph convolution with a fusion module guided by mutual information maximization. By incorporating diverse views of spatial data and aligning their representations, MCFST effectively captures latent patterns and achieves robust domain recognition. We evaluated MCFST against state-of-the-art methods on two simulated datasets with varying sparsity and noise levels, as well as three real spatial transcriptomics datasets. Results show that MCFST consistently outperforms baselines in spatial domain identification, highlighting its robustness and efficiency. Moreover, spatially variable genes detected from MCFST-derived domains exhibited clear spatial expression patterns, further confirming the accuracy and utility of MCFST. AVAILABILITY: The code implementation of the MCFST algorithm is publicly available at https://github.com/dw666666/MCFST. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
SpaBiT: Enhancing Spatial Transcriptomics Resolution via Bidirectional Attention Transformers.
Liu X, Li A, Min W
Bioinformatics · 2026 Jul · PMID 42391614 · Publisher ↗
MOTIVATION: Spatial transcriptomics (ST) enables the precise mapping of gene expression within tissue architecture, however its application is often limited by low spatial resolution and sparse sampling. While existing d...
MOTIVATION: Spatial transcriptomics (ST) enables the precise mapping of gene expression within tissue architecture, however its application is often limited by low spatial resolution and sparse sampling. While existing deep learning methods leverage histology images, spatial coordinates, or low-resolution expression data to predict high-density profiles, these methods are limited in either capturing the intrinsic constraints between histological context and spatial topology or ignoring the complex local neighborhood relationships between spots. RESULT: To address these limitations, we propose SpaBiT, a multimodal framework designed to enhance ST resolution via a bidirectional attention mechanism. At its core, SpaBiT employs a bidirectional cross-attention module to facilitate precise information exchange between image features and neighborhood-aware representations learned via a graph attention network. This design explicitly models the synergistic constraints between local morphology and spatial graph topology, yielding high-fidelity, high-density gene expression maps. SpaBiT exhibits competitive performance in reconstructing complex spatial gene expression, outperforming the benchmark models utilized in this study across various quantitative metrics, providing a robust tool for deciphering complex tissue microenvironments. AVAILABILITY: The source code and datasets are available at https://github.com/wenwenmin/SpaBiT. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
EDEL: Enhancing Dense Retrievers for Curation of Biomedical Knowledge Bases.
Wang XD, Leser U
Bioinformatics · 2026 Jul · PMID 42391609 · Publisher ↗
MOTIVATION: Retrieval of relevant papers from the literature is the first step in curating high-quality biomedical knowledge bases. While BM25 has long been the method of choice, dense retrieval models achieved improved...
MOTIVATION: Retrieval of relevant papers from the literature is the first step in curating high-quality biomedical knowledge bases. While BM25 has long been the method of choice, dense retrieval models achieved improved accuracy by embedding queries and documents into dense vector representations. Existing knowledge bases provide a natural source for deriving query-document pairs for training such models. Current training approaches, however, do not take into account that some evidence described by knowledge base entries may only be partially expressed in document abstracts, while the full evidence is often contained in inaccessible full texts, introducing noise into binary relevance labels. In addition, existing approaches only make limited use of the knowledge base structure for selecting negative samples during training. RESULTS: We propose EDEL, a novel dense bi-encoder for biomedical knowledge base curation to enable curators to find relevant papers for annotation faster. It introduces a loss function using graded relevance scores instead of binary labels to facilitate learning from partially grounded examples, together with a structured sampling strategy that exposes the model to diverse and hard negative examples during training. We evaluate EDEL's performance in two curation settings, namely precision oncology (on CIViC and OncoKB) and post-translational modifications (on UniProt). EDEL outperforms other state-of-the-art models in NDCG@10 by 1.5 and 3.4 percentage points, respectively. Ablation studies show the effectiveness of both innovations. These results indicate that EDEL can substantially improve literature retrieval for biomedical knowledge base curation. AVAILABILITY AND IMPLEMENTATION: Code to reproduce our results is available at: https://github.com/WangXII/edel_repo. SUPPLEMENTARY INFORMATION: Supplementary data is attached to this manuscript.
Informative Relational Learning for Adverse Reaction Prediction with Enhanced Generalization to Novel Drugs.
Sun S, Zhang D, Chu H … +1 more , Gong X
Bioinformatics · 2026 Jul · PMID 42391027 · Publisher ↗
MOTIVATION: Accurate prediction of adverse drug reactions (ADRs) is essential for drug safety surveillance, and recent advances in machine learning with heterogeneous biomedical information have improved predictive perfo...
MOTIVATION: Accurate prediction of adverse drug reactions (ADRs) is essential for drug safety surveillance, and recent advances in machine learning with heterogeneous biomedical information have improved predictive performance. However, two challenges remain: current methods often learn inadequate ADR representations that fail to capture dependencies among ADRs, and generalize poorly to novel drugs. RESULTS: To obtain informative ADR embeddings, we construct a multi-source, multi-relational ADR graph that integrates hierarchical structure and empirical ADR co-occurrence, and apply a relational graph convolutional network (R-GCN) to learn relation-aware ADR representations. To enhance generalization to novel drugs, we exploit the hierarchical structure of the Anatomical Therapeutic Chemical (ATC) classification to link drugs via shared higher-level categories for effective knowledge transfer and model these relations with an R-GCN. We further introduce a Conditional Domain Adversarial Network (CDAN) to reduce distribution shifts between known and novel drugs by aligning features conditioned on predicted ADR labels, learning domain-invariant yet task-relevant representations. Additionally, to exploit similar ADR patterns among related drugs, we introduce a dual-branch mixture-of-experts (Dual-MoE) module where each expert captures ADR commonalities within a drug category in one branch, while a separate branch models global patterns. Extensive experiments show that our method consistently outperforms seven baselines, achieving F1 improvements of 4.3% and 4.7% over the best baseline on two datasets, respectively, with more balanced precision-recall trade-offs. It also improves AUC on uncommon ADRs by 7% more than on common ADRs, and remains more robust under data sparsity, with more gradual performance degradation as training data decreases. AVAILABILITY AND IMPLEMENTATION: The code of our model is available at https://github.com/fzsdb/Knowledge-guided-ADR-prediction.git.
An interpretable deep learning framework uncovers features governing CRISPR-Cas9 genome-editing efficiency.
Bakhtiyari N, Masoudi-Sobhanzadeh Y, Farajnia S … +1 more , Kumar S
Bioinformatics · 2026 Jul · PMID 42391025 · Publisher ↗
MOTIVATION: CRISPR-Cas9 genome-editing efficiency is strongly influenced by the sequence composition and positional context of single-guide RNAs (sgRNAs). Although numerous deep learning-based models have been developed...
MOTIVATION: CRISPR-Cas9 genome-editing efficiency is strongly influenced by the sequence composition and positional context of single-guide RNAs (sgRNAs). Although numerous deep learning-based models have been developed to predict Cas9 efficiency from sgRNA sequences, most operate as black boxes, offering limited insight into the sequence determinants underlying Cas9 activity. In addition, previous studies often overlook how the positional context of sequence motifs within sgRNAs influences their effects on Cas9 binding or cleavage. RESULTS: We introduce DeepCC9, an interpretable machine learning framework that combines explicit sequence feature extraction with a residual block-based deep architecture to improve interpretability and identify composition- and position-based motifs governing Cas9 genome-editing efficiency. We applied this method to multiple Cas9 variant datasets, achieving superior predictive performance compared with existing methods while enabling direct interpretation of sequence motifs and their positional effects. Our analysis uncovered 74 sequence motifs enriched or depleted at specific positions within sgRNAs and strongly associated with Cas9 efficiency, providing mechanistic insight into sequence features that influence guide performance. Together, these results establish DeepCC9 as a generalizable and interpretable framework for modeling sequence-function relationships and advancing the understanding of the sequence determinants underlying CRISPR-Cas9 genome editing. AVAILABILITY AND IMPLEMENTATION: The authors have implemented their algorithm in the Python programming language (version 3.X), which is accessible using (https://github.com/MasoudiYosef/DeepCC9, https://zenodo.org/records/20073890). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
3DICE: Interpretable 3D Cross-Modal Learning for Drug-Target Interaction Prediction and Large-Scale Drug Discovery.
Liu AZR, Le NQK, Chua MCH
Bioinformatics · 2026 Jul · PMID 42390120 · Publisher ↗
MOTIVATION: Drug-target interaction (DTI) prediction is a crucial step in modern drug discovery. Accurate and efficient predictions can substantially reduce costs and development time. Applications of deep learning metho...
MOTIVATION: Drug-target interaction (DTI) prediction is a crucial step in modern drug discovery. Accurate and efficient predictions can substantially reduce costs and development time. Applications of deep learning methods for this purpose have been extensively studied in recent years, yielding instrumental contributions to this field. However, existing methods face issues pertaining to efficient learning of drug and target feature representations, which is detrimental to generalisability and performance in cold-start scenarios. Most approaches extract representations from SMILES strings for drugs and FASTA sequences for target proteins, which encode limited 3D structural information. Additionally, many models lack explainability, being black boxes that provide little physical insight into the underlying mechanisms behind such interactions. RESULTS: We propose 3DICE, a novel framework leveraging co-attention-based fusion and massively pre-trained 3D structural encoders for both drugs and proteins. Uni-Mol and ESM-IF1 are employed to generate high-fidelity, 3D structure-aware embeddings which enable richer geometric and chemical understanding. Cross-modal fusion modules further augment representations to model intermolecular binding relationships. Importantly, this mechanism also provides intrinsic interpretability, highlighting and enabling qualitative analysis of most influential atoms or residues. Experiments conducted on two canonical benchmark datasets display the competitiveness of our model in real-world scenarios. 3DICE outperformed state-of-the-art models across multiple metrics on the DrugBank and KIBA datasets. Additional experiments provide a more rigorous analysis of interpretability than is typically reported in prior DTI studies, and we find that attention consistently highlights decision-critical regions which is not intrinsically class-specific. AVAILABILITY: Our model and dataset are freely available at: https://github.com/austinatose/3DICE. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
KASSPer: Kinase Active Site Structure Prediction using Protein and Ligand Language Models and Its Application to Virtual Screening.
Jang W, Shin WH
Bioinformatics · 2026 Jul · PMID 42386696 · Publisher ↗
MOTIVATION: Structure-based virtual screening (SBVS) is limited by the rigid-receptor assumption, which is particularly problematic for kinases that adopt multiple active-site conformations but are experimentally biased...
MOTIVATION: Structure-based virtual screening (SBVS) is limited by the rigid-receptor assumption, which is particularly problematic for kinases that adopt multiple active-site conformations but are experimentally biased toward a single state. Although ensemble screening can address this limitation, it remains computationally expensive. RESULTS: We introduce KASSPer (Kinase Active Site Structure Predictor), a framework that predicts kinase active-site conformational states using protein and compound language models. Given a kinase amino acid sequence and a ligand SMILES string, KASSPer enables ligand-specific conformer selection prior to SBVS, potentially reducing the computational cost associated with exhaustive ensemble screening. Benchmarking on the DUD-E kinase subset demonstrates that KASSPer-guided screening outperforms the tested ensemble-based approach across the evaluation metrics. AVAILABILITY AND IMPLEMENTATION: The implementation for model loading and inference is available at the GitHub repository https://github.com/kucm-lsbi/KASSPer. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
IDR searcher: a search engine solution for public image resources.
Mohamed K, Moore W, Lindner D … +5 more , Moore J, Swedlow JR, Walczysko P, Wong F, Burel JM
Bioinformatics · 2026 Jul · PMID 42386687 · Publisher ↗
MOTIVATION: In recent years, public image resources have emerged, but finding quality data efficiently remains a challenge, therefore limiting reuse. RESULTS: IDR searcher is an open-source search engine designed to faci...
MOTIVATION: In recent years, public image resources have emerged, but finding quality data efficiently remains a challenge, therefore limiting reuse. RESULTS: IDR searcher is an open-source search engine designed to facilitate the exploration of datasets hosted in public bioimaging resources. The application offers a fast, efficient, cost-effective solution for discovering datasets and has the potential to address current disparities in finding quality datasets for exploratory research and can be combined with metadata visualization tools to enhance usability for the scientific community. AVAILABILITY: IDR searcher is deployed using Ansible playbooks and released under the GPL v2 license. The source code associated with this manuscript is available at https://doi.org/10.5281/zenodo.20641515. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
KCFtools: rapid alignment-free method for introgression screening and GWAS using k-mer profiles.
Selvanayagam S, Quiroz-Chavez J, Ramirez-Gonzalez RH … +3 more , Uauy C, Smit S, Schranz ME
Bioinformatics · 2026 Jul · PMID 42384922 · Full text
MOTIVATION: In the era of multiple genome references, researchers often align sequencing reads against distinct assemblies or even multiple references simultaneously. This enables applications such as the detection of in...
MOTIVATION: In the era of multiple genome references, researchers often align sequencing reads against distinct assemblies or even multiple references simultaneously. This enables applications such as the detection of introgressed segments or highly variable genomic regions, which are especially prevalent in large-genome crop species such as lettuce or wheat. However, these applications come at the cost of increased computational burden, inconsistencies in mapping methods, and reduced reproducibility across studies. To address these limitations, we developed KCFtools, a Java-based toolkit that identifies the presence and absence of k-mers in nonoverlapping genomic or transcriptomic windows by comparing query and reference genomes. This alignment-free approach enables the efficient computation of an identity score for each window, thereby facilitating robust detection of introgressed or variable regions across genomes. RESULTS: We systematically evaluated the performance and accuracy of the k-mer-based method implemented in KCFtools, benchmarking it against conventional single nucleotide variation-based introgression detection pipelines. Our results demonstrate that KCFtools effectively captures introgressed segments and structurally diverse regions, even in species with fragmented or highly divergent reference genomes. In addition, we extended KCFtools to generate genotype matrices from k-mer variation tables. These matrices are compatible with genome-wide association studies software and allow the identification of loci associated with phenotypic traits. We showcase the utility of this approach by detecting known and novel associations for downy mildew resistance in lettuce, underscoring the pipeline's potential for high-resolution, reference-agnostic population genetic analysis. AVAILABILITY AND IMPLEMENTATION: https://github.com/sivasubramanics/kcftools.
Meta2DB: Curated shotgun metagenomic feature sets and metadata for health state prediction.
Kok CR, Mulakken NJ, Thissen JB … +8 more , Martí JM, Lee R, Trainer JB, Goncalves AR, Ranganathan H, Avila-Herrera A, Jaing CJ, Be NA
Bioinformatics · 2026 Jul · PMID 42384916 · Publisher ↗
SUMMARY: Meta2DB is a curated metagenomic and metadata database that provides structurally consistent microbiome taxonomy feature count tables for 13,897 samples across 84 studies, 23 disease states, and 34 geographical...
SUMMARY: Meta2DB is a curated metagenomic and metadata database that provides structurally consistent microbiome taxonomy feature count tables for 13,897 samples across 84 studies, 23 disease states, and 34 geographical locations. All samples were uniformly processed using a streamlined metagenomic classification pipeline that employs a unique and comprehensive reference database indexed to contain all sequences across all kingdoms of life that were present in the NCBI Nucleotide (nt) database retrieved on January 04, 2023. This pipeline leverages high-performance computing (HPC) resources at Lawrence Livermore National Laboratory and was used to process 50TB of publicly available raw metagenomic sequence data. Extensive metadata curation was carried out through a combination of manual curation and automated parsing, producing a consistent inter-study metadata table specifically structured to facilitate training of ML models for prediction of human health. AVAILABILITY: Data is available at https://gdo-meta2db.llnl.gov/ and https://zenodo.org/records/17315984. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
conMItion: an R package adjusting confounding factors for associations in multi-omics.
Wang G, Liu F, Chen Z … +1 more , Davoli T
Bioinformatics · 2026 Jun · PMID 42378451 · Publisher ↗
SUMMARY: Association measurements, such as mutual information (MI), are fundamental in the analysis of cancer multi-omics data for identifying cancer-related genes, gene signatures, and gene regulatory networks, thereby...
SUMMARY: Association measurements, such as mutual information (MI), are fundamental in the analysis of cancer multi-omics data for identifying cancer-related genes, gene signatures, and gene regulatory networks, thereby shedding light on tumor development, progression, and treatment. Confounding factors, including tumor purity and mutation burden, can bias association measurements in MI, potentially leading to the misclassification of passenger events as drivers. Conditional mutual information (CMI) provides a robust framework for assessing both linear and nonlinear associations while effectively accounting for different confounding factors. An R package called conMItion is introduced to estimate CMI and its statistical significance for multi-omics data, with the flexibility to adjust for one or two confounding factors. We demonstrated the utilization of conMItion through two use cases. First, we identified interchromosomal somatic copy number alteration-expression associations in bladder cancer. Second, we identified associated cell types within the lung cancer tumor microenvironment using single-cell RNA sequencing datasets. AVAILABILITY AND IMPLEMENTATION: The conMItion package is freely available on CRAN at https://CRAN.R-project.org/package=conMItion. The two use cases described in the paper can be accessed at https://github.com/GJYWang/conMItion. A supplementary document is available online.
SpaMFG: a Spatial Multi-omics Integration Method based on Feature Grouping.
Li Z, Ma L, Liu J … +4 more , Sun W, Li Y, Zhao C, Yu L
Bioinformatics · 2026 Jun · PMID 42378450 · Publisher ↗
MOTIVATION: The rapid development of spatial multi-omics technology enables the simultaneous measurement of gene and protein expression alongside spatial location, providing valuable insights into tissue heterogeneity. H...
MOTIVATION: The rapid development of spatial multi-omics technology enables the simultaneous measurement of gene and protein expression alongside spatial location, providing valuable insights into tissue heterogeneity. However, challenges such as low spatial resolution and high feature dimensionality complicate data integration and biological interpretation. RESULTS: To address these issues, we propose SpaMFG, an innovative feature-group-level framework for interpretable spatial multi-omics integration. SpaMFG leverages spatial location information and introduces a spatial proximity weighting method to improve feature grouping accuracy. Additionally, it employs a new cross-omics feature group matching method that combines spatial location and Jaccard similarity to construct a weighted cost matrix, which is optimized using the Hungarian algorithm. This approach enhances the biological interpretability of cross-omics feature relationships. We evaluated SpaMFG's performance through comparative analysis on the human lymph node dataset, demonstrating its effectiveness. Further applications on human tonsils, mouse spleens, and mouse thymus datasets confirmed the robustness of SpaMFG in various biological contexts. AVAILABILITY: The source code for SpaMFG is available at https://github.com/LiangYu-Xidian/SpaMFG. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
CSCN: Inference of Cell-Specific Causal Networks Using Single-Cell RNA-Seq Data.
Wang M, Yang J, Lyu L … +1 more , Chen J
Bioinformatics · 2026 Jun · PMID 42378449 · Publisher ↗
MOTIVATION: Understanding gene regulation is fundamental to deciphering the coordinated activity of genes within cells. Although single-cell RNA sequencing (scRNA-seq) enables gene expression profiling at cellular resolu...
MOTIVATION: Understanding gene regulation is fundamental to deciphering the coordinated activity of genes within cells. Although single-cell RNA sequencing (scRNA-seq) enables gene expression profiling at cellular resolution, most gene network inference methods operate at the tissue or population level, thereby overlooking regulatory heterogeneity across individual cells. Recent approaches, such as Cell-Specific Network (CSN) and its extension c-CSN, attempt to construct gene networks at single-cell resolution, providing a more detailed view of the regulatory logic underlying individual cellular states. However, these methods remain limited by high false positive rates due to indirect associations and lack of directionality or causal interpretability. RESULTS: To address these issues, we propose the Cell-Specific Causal Network (CSCN) framework, which infers directed, cell-specific gene regulatory relationships by explicitly modeling causality. CSCN combines causal discovery techniques with efficient computation using kd-trees and bitmap indexing to perform conditional independence testing, yielding sparse and interpretable causal graphs for each cell that effectively suppress indirect and spurious associations. Across nine scRNA-seq datasets, the Causal Katz Matrix (CKM) derived from CSCN provided more accurate and stable cell-state discrimination than expression-based and network-based baselines. CSCN-derived representations also preserved developmental structure, achieving the best trajectory performance in simulations and the strongest agreement with human embryo progression. Beyond RNA-only analysis, CSCN further generalized to paired PBMC multiome, CITE-seq, and spatial transcriptomic settings. Also, in controlled confounding simulations, CSCN consistently achieved the lowest false-positive rates relative to CSN and c-CSN. AVAILABILITY AND IMPLEMENTATION: The code is available at https://github.com/open17/CSCN. SUPPLEMENTARY INFORMATION: Supplementary data are available online.
Sparse CCA-Based Mediation Analysis with High-Dimensional Exposures and Mediators.
Li X, Kong M, Smith MR … +17 more , Liang Y, Teeny S, Ly VT, Go YM, Samala N, Jones DP, Luo J, Watson WH, McClain CJ, Vatsalya V, Szabo G, Dasarathy S, Mitchell M, Nagy L, Barton B, Cave MC, Jiang H
Bioinformatics · 2026 Jun · PMID 42378448 · Publisher ↗
MOTIVATION: Mediation analysis plays a crucial role in understanding how exposure variables influence health outcomes via intermediate variables, or mediators, in environmental studies. When analyzing a large number of e...
MOTIVATION: Mediation analysis plays a crucial role in understanding how exposure variables influence health outcomes via intermediate variables, or mediators, in environmental studies. When analyzing a large number of environmental exposures, such as chemical mixtures or pollutants, together with multiple potential mediators such as metabolites, advanced methodologies are necessary to accurately separate direct and indirect effects. This paper proposes a novel mediation analysis method based on Sparse Canonical Correlation Analysis (SCCA), designed specifically for settings where both exposures and mediators are high-dimensional. The effectiveness of the proposed method is evaluated through simulation studies and an application to real-world data. RESULTS: The proposed SCCA-based mediation framework improved identification of relevant mediators and pathways in simulation studies, particularly in high-dimensional and noisy settings. The two-step screening extension further enhanced feature selection while maintaining stable estimation. In the real-data application, the method identified interpretable exposure-metabolite pathways associated with MELD score, with several pathways showing moderate selection stability and robustness to potential unmeasured confounding. AVAILABILITY: The R code for implementing the proposed method and the simulation studies is available at https://github.com/MaggieLi2001/HDM-SCCA2. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
Enhancing Cross-Context Generalization in Drug Perturbation Prediction with a Multimodal Conditional Diffusion Framework.
Ma Y, Du K, Li Y … +2 more , Li P, Yu L
Bioinformatics · 2026 Jun · PMID 42378442 · Publisher ↗
MOTIVATION: Predicting drug-induced transcriptional perturbations is critical for precision medicine, yet existing models fail to capture multimodal biological context, limiting generalization across unseen drugs and cel...
MOTIVATION: Predicting drug-induced transcriptional perturbations is critical for precision medicine, yet existing models fail to capture multimodal biological context, limiting generalization across unseen drugs and cell lines. RESULTS: We present PertDiff, a conditional diffusion framework that integrates control gene expression, LLM-derived cell semantics, and pretrained molecular graph representations to predict transcriptome-wide perturbations. PertDiff outperforms state-of-the-art baselines in prediction accuracy and generalizes robustly across drugs and cell lines. It further demonstrates translational utility through accurate drug sensitivity prediction, therapeutic repurposing for pancreatic cancer, and concordance with real-world clinical treatment outcomes, establishing it as a biologically grounded transcriptomic modeling tool. AVAILABILITY: The source code and data are available at https://github.com/Panda-myj/PertDiff and https://doi.org/10.5281/zenodo.18427848. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
Primer Design through Submodular Function Estimation.
Chen Y, Han Y, Wang A … +4 more , Hong A, Rivers AR, Kuhnle A, Boucher C
Bioinformatics · 2026 Jun · PMID 42378440 · Publisher ↗
MOTIVATION: Multiplex PCR-based enrichment is widely used in viral genome sequencing and pathogen surveillance. However, designing large sets of primers that maximize genome coverage while minimizing primer-primer intera...
MOTIVATION: Multiplex PCR-based enrichment is widely used in viral genome sequencing and pathogen surveillance. However, designing large sets of primers that maximize genome coverage while minimizing primer-primer interactions remains a major computational challenge. Existing methods such as SADDLE and Olivar use heuristics to optimize a Badness score for primer dimers but lack theoretical guarantees on solution quality. RESULTS: We introduce PRISM, a new framework that formulates multiplex primer design as a constrained submodular maximization problem. Our method defines an objective that balances genome coverage and dimer risk, and applies a local search algorithm with a constant-factor approximation guarantee. Evaluations on viral genome datasets demonstrate that PRISM consistently achieves lower Badness scores compared to PrimalScheme, Olivar, and primerJinn. These results highlight the scalability and theoretical rigor of submodular optimization in primer design. AVAILABILITY: PRISM is open-source and available at https://github.com/yhhan19/PRISM-new. The experimental data, scripts, and results used in this paper are archived on Figshare at https://doi.org/10.6084/m9.figshare.32806499. SUPPLEMENTARY INFORMATION: Supplementary information: Supplementary material, including proofs and figures, is available at Bioinformatics online and on Figshare at https://doi.org/10.6084/m9.figshare.32806499.
About
- Frequency
- Sun
- Papers found
- 200
- RSS feed
- Subscribe