BioData Mining

Background: Genetic conditions are common in congenital heart disease (CHD) and affect care, but are often missed or diagnosed late. We asked whether routine clinical data can shortlist the most likely genetic conditions. Methods: Retrospective single-center cohort of children with CHD and an echocardiogram who had either a confirmed genetic diagnosis or negative cytogenetic testing and CHD gene sequencing. Echocardiogram images and serial medical record data were added as model inputs in the order they become clinically available. Class balanced logistic regression (LR) and a multimodal deep learning fusion network were fitted to the same inputs at each step. The primary metric was balanced top-3 accuracy (correct diagnosis among the top three, averaged across conditions), with bootstrap intervals. Results: The cohort included 1,720 children (988 with common CHD-associated genetic conditions, 231 rare diagnoses, 501 negative testing). Using only inputs generated at the echocardiogram visit, LR reached a balanced top-3 accuracy 0.758 (chance 0.429). Serial records, radiology notes and ICD-10 codes, raised this to 0.809. Multimodal fusion performed similarly at each stage. With a genetic condition withheld from training, the fusion embedding placed affected children nearest to others sharing a gene group in 0.37 of cases versus 0.10 without the embedding (+0.268 [95% CI +0.242, +0.295]). Conclusions: Routine clinical data available at first echocardiogram can help identify likely genetic diagnoses, and later coded inputs ceiling. Multimodal fusion added no accuracy over LR, but places children sharing a genetic cause near one another, making nearest-neighbor retrieval informative for undiagnosed children.

In recent years, the issue of reproducibility in scientific studies has been at the forefront of scientific discussion. Many of these issues arise due to poor employment of statistical methods, making it crucial to address these knowledge gaps for frequently used methods. Elastic net regression (ENET) is one such algorithm used frequently in biology and public health due to its ability to select important predictors (also known as "features") from large data sets. However, ENET struggles with consistent feature selection and hyperparameter choice in these applications due to low observation counts and noisy data. This limitation hinders ENET's effectiveness in biological research due to low model reproducibility and predictor interpretability. We propose the Instability Quotient Criterion (IQC) as a metric to quantify the consistency of selected features for feature selection models such as ENET, with a specific emphasis on biological applications. IQC leverages ensemble modeling and Shannon's entropy in order to determine the stability of selected features derived from ENET models. We demonstrate IQC's robustness via simulation studies, showing that IQC is capable of detecting significant decreases in model stability with lower observation counts, higher dimensionality, and higher proportions of noisy signal. We additionally utilize simulation to provide a naive threshold for low versus high stability. Finally, we applied IQC to a case study to illustrate how our method can be integrated into a larger analysis pipeline when looking at microarray gene expression data. IQC indicated stability issues for differentially-expressed genes when classifying acute leukemia subtypes in spite of high classification success. Lastly, we discuss how model instability impedes interpretation of biologically significant genes for discerning cancer subtypes. Overall, we present IQC as a robust evaluation metric that can easily be incorporated into high-dimensional biological analyses.

When an ontology-based biomedical knowledge graph (a KG, containing information about genes, pathways, phenotypes, and molecular interactions) is analyzed using a computational reasoning system, the system can return unwanted generic concepts. Such generic concepts, in the context of a query about potential drug treatments for a disease, might include 'pharmaceutical preparations' or 'antibiotics'. Answers can be filtered based on hand-curated block-lists or ontology Information Content (IC), but to our knowledge, (1) neither filtering method has been validated against expert judgments of generality, and (2) IC is not available for the majority of nodes in the representative KG used in this work. In this study, three domain experts rated representative lists of KG concepts for generality. We built a semantic precision score (SPS) from a concept's name and description alone, using a biomedical language model (BioLORD). For the purpose of filtering generic answers from biomedical query results, we compared (using queries for 200 diseases) the SPS against a curated concept block-list and two IC thresholds. IC agreed moderately with the experts (Spearman's {rho} = 0.59), but correlations were negligible or negative within IC bands. The SPS was well-correlated with the experts ({rho} = 0.76) globally and within SPS bands, and remained informative on concepts without IC ({rho} = 0.59). Hard filtering of query results using the SPS removed 1,502 of 33,161 query answers, nearly three times the block-list's 513, while removing fewer of the query-specific DrugCentral indications. Hard filtering by IC removed either few answers or many approved drugs. We additionally compared the filtering methods' performance on an independent evaluation set of queries for 200 diseases. The SPS filter removed 5,135 of 91,203 answers compared with the block-list's 1,419 and lost none of 1,296 indications compared with 17. Further, SPS filtering removed approved drugs no more often from the half of DrugCentral not used in training than from the half used in training. Semantic meaning captures expert judgments of specificity more faithfully than ontology position, and semantic scoring can replace IC and manually maintained block-lists for filtering drug-repurposing results.

Integrating heterogeneous molecular and clinical data into unified, clinically meaningful patient representations remains a major challenge in precision medicine. Here, we present SurvMOCA-GNN, a biologically informed multi-omics integration framework that explicitly incorporates biological network structure, cross-omics regulatory relationships, and patient-specific clinical information to generate unified patient representations. Using gene expression, microRNA expression, and clinical data from the TCGA-LAML cohort, SurvMOCA-GNN revealed patient representations that identified prognostically distinct patient groups and further stratified patients within established European LeukemiaNet 2022 risk categories. Systematic ablation analyses demonstrated that incorporation of graph-based modeling, cross-omics attention, adaptive clinical integration, and biological prior knowledge progressively improved integration quality, resulting in more structured patient representations and stronger prognostic stratification. Compared with existing multi-omics integration approaches, SurvMOCA-GNN consistently produced more biologically coherent patient representations with improved survival discrimination. Evaluation in an independent AML cohort, together with transfer learning analyses, further demonstrated the robustness and transferability of the integrated representations across heterogeneous patient populations. Together, these findings demonstrate that biologically informed multi-omics integration reveals clinically meaningful patient representations and provides a general framework for integrating heterogeneous molecular and clinical data to improve patient stratification in AML and potentially other diseases. The SurvMOCA-GNN framework is freely available at https://github.com/tjgu/SurvMOCA-GNN.git.

Background: Epigenetic deregulation is a defining feature of leukemia, and DNA methylation and chromatin accessibility profiles encode the cell of origin, lineage and clinical behaviour of the disease. Artificial intelligence (AI) and machine learning (ML) can extract predictive patterns from these high-dimensional data, but the published models have not been systematically appraised. We systematically reviewed AI models built on epigenomic data in leukemia for diagnosis, prognosis, relapse and treatment-response prediction, and measurable residual disease (MRD) detection. Methods: This review followed PRISMA 2020 and the SWiM guideline. PubMed was searched on 7 October 2026 with three structured queries, supplemented by a broader PubMed query, citation searching and web searching. Eligible studies were peer-reviewed primary studies that applied an ML or deep-learning algorithm to DNA methylation, chromatin-accessibility or histone-modification data from human leukemia samples and reported a quantitative diagnostic, prognostic, response or MRD outcome. Risk of bias and applicability were assessed with PROBAST, informed by PROBAST+AI. Owing to heterogeneity, results were synthesized narratively and in structured tables, without meta-analysis. Results: Of 68 unique database records and 32 reports from other sources, 30 studies (2010-2026; 22 since 2020) were included: 16 in acute myeloid leukemia, 4 in acute lymphoblastic leukemia, 2 spanning acute leukemia lineages, 3 in juvenile myelomonocytic leukemia and 5 in chronic lymphocytic leukemia; 10 were pediatric. DNA methylation was the input in 28 studies and chromatin accessibility in 2; no eligible study modelled histone marks. Tree ensembles (n = 9) and support vector machines (n = 5) were the most frequent algorithms. Diagnostic classifiers reported accuracy, concordance or F1 scores of 0.87-1.00, and two classifiers were applied to nanopore sequencing data, one returning acute leukemia subtype calls within about 2 hours. Prognostic epigenomic scores were associated with survival (reported hazard ratios 2.36-10.82), but discrimination fell on transport to independent cohorts (C-index 0.677 internally vs 0.529 in one external cohort). Fifteen studies validated models on an external epigenomic cohort, one prospectively. Overall risk of bias was high in 21 studies, unclear in 8 and low in 1, driven by the analysis domain (small samples, absent calibration, possible data leakage). Calibration was reported in 3 studies and code was available for 11. Conclusions: Epigenomic AI models classify leukemia subtypes with high reported accuracy and can deliver same-day diagnoses, and methylation-based scores add prognostic information. However, the evidence is dominated by retrospective, high-risk-of-bias development studies with limited external validation and almost no prospective or impact evaluation. Prospective, multi-center validation with calibrated, transparently reported models is required before clinical adoption.

Motivation: Traditional methods primarily rely on statistical features such as k-mers and GC content, making it difficult to capture complex internal relationships within sequences. Deep learning models typically rely on discrete encodings, leading to issues such as information sparsity, dimensional redun-dancy, and disruption of sequence continuity. Our previous UniWave-1 framework transforms se-quences into waveform signals; however, it lacks comprehensive physicochemical integration and com-putational efficiency. Results: We propose the UniWave-2 feature extraction framework. UniWave-2 incorporates three key advancements: (i) integration of hydrophobicity into the encoding scheme; (ii) enhancement of wave-form resolution through Fourier transform, coupled with wavelet transform to precisely capture both local details and global periodic patterns; and (iii) development of a lightweight multi-scale GRU model that facilitates cross-dimensional feature interactions, captures bidirectional temporal dependencies, and integrates long- and short-range patterns with spatial positional awareness, thereby yielding rich feature representations. UniWave-2 achieved competitive performance across multiple genomic tasks, while ISM analysis demonstrated that waveform encoding enables interpretable attribution without re-lying on gradient information. These results establish UniWave-2 as a lightweight and interpretable framework for nucleic acid sequence analysis.

Background and Objective: Prescribing multiple drugs simultaneously, known as polypharmacy, is increasingly common in managing chronic diseases such as diabetes and cardiovascular conditions. Studies show that roughly 40% of elderly patients are on five or more medications, putting them at significant risk of adverse drug-drug interactions (DDIs) [1]. While computational methods based on Graph Neural Networks (GNNs) have shown strong potential for predicting these interactions from a drugs molecular structure, most existing models are limited in two important ways: they rely on attention mechanisms that cannot adapt to the unique context of each atom in a molecule, and they only predict whether an interaction exists rather than how dangerous it might be. Methods: We propose HMVA-DDI, a two-stage deep learning framework that addresses both shortcomings. In the first stage, our model predicts whether two drugs will interact. If it detects an interaction, the second stage classifies it into one of three clinical severity levels: minor, moderate, or major. To better capture molecular structure, we replace the standard attention mechanism with a more powerful version called Graph Attention Network v2 (GATv2), which considers each atoms specific context when learning its representation. We also introduce a technique to prevent information loss in deeper network layers, and we tailor the input features separately for each prediction task. Results: Evaluated on two standard benchmarks, DrugBank 5.0.3 [2] and DDInter [3], HMVA-DDI achieves an AUROC of 0.9819 and AUPRC of 0.9759 on binary interaction detection, a marked improvement over the baseline. For severity classification, our best model variant (qcSvr4) reaches an overall AUROC of 0.9444, with a particularly large improvement in detecting low-severity (Minor) interactions, a historically difficult class. Conclusion: HMVA-DDI bridges the gap between computational drug research and clinical usefulness by not only predicting interactions accurately but also indicating interaction severity. This two-level system can help healthcare professionals prioritize warnings, reducing the alert fatigue problem where doctors are overwhelmed by too many non-critical notifications.

Tabular synthetic data generation (SDG) can facilitate research and development in the health domain, as access to health data is restricted. To accommodate for its structural complexity, such as multimodal distributions and fragmentation, transformer models provides an alternative to conventional SDG methods since they are better equipped to infer global spatial coherence due to textual context of tabular records. However, methods for optimising their adaption to tabular SDG are still largely unexplored. We contribute to the understanding of transformer-based SDG by evaluating fidelity and privacy of three prompting regimes, zero shot, one shot and few shot training, for tabular health SDG using small transformer models. We evaluated llama 3.2 (1B and 3B parameters) and Phi (1.3B and 2.7B parameters), on a real clinical dataset. Our results showed that zero shot learning was less prone to hallucinations and performed overall better for fidelity, but at the loss of privacy in comparison to the other regimes. Our results showed no differences between model sizes and families. Future studies should extend the analysis to other models and datasets, and explore strategies to reduce hallucinations.

Quantitative characterization of cell proliferation is central to preclinical drug discovery. Here, we evaluated Random Forest (RF) regression for predicting confluence-based cell growth trends using data from human cancer cell lines and benchmarked its performance against the widely used logistic and Gompertz models. The RF model achieved higher predictive accuracy within the evaluated dataset, particularly at higher initial seeding densities. We further extended the RF framework to predict growth curves across seeding densities. Collectively, these findings highlight the potential of data-driven modeling to complement conventional mathematical approaches and support more informed cell-culture design.

Bioinformatics software plays a central role in biological research, yet many academic projects lack robust testing and continuous integration (CI) practices. This absence can compromise reproducibility, accessibility, and long-term maintainability. In this study, we conduct an empirical analysis of GitHub repositories spanning bioinformatics and other scientific software domains to examine the relationship between CI, software testing, and software quality. Repository discovery was performed using domain-specific search terms via GitHub's Application Programming Interface (API). This was followed by a layered data extraction pipeline, which included a Large Language Model (LLM) validation step, that connected repositories to published works, institutions, authors, funding, and citation counts. By integrating repository activity, community engagement, and backlog health into a composite Repository Quality Index (RQI), we find that the presence of CI and software tests is significantly associated with higher repository quality. Additionally, repositories with higher academic citation counts tend to cluster in higher RQI tiers, suggesting that software quality, which is driven by proper testing and integration, is associated with academic impact and funding status. These results demonstrate that adopting industry-standard development practices, such as CI and software testing, is a tractable path toward more sustainable and impactful academic software and funding opportunities.

Model averaging combines the predictions of several population pharmacokinetic models to support model-informed precision dosing, but how to select ensemble models remains unexplored. Using tacrolimus as a case study, we first evaluated all possible ensembles of 27 models in 37 kidney and liver transplant recipients with full pharmacokinetic profiles at occasions 1 and 5 and trough concentrations at occasions 2, 3 and 4. Models were weighted by individual objective function value. Predictive performance for the occasion 5 area under the curve was assessed with individual parameters estimated from the occasion 5 trough concentration or from occasion 1 to 4 observations. Then, seven ensembles, selected a priori from relevance and external evaluation scores or as the best-performing ensembles previously identified, were compared across ten sets of observations, from a single trough concentration to dense sampling. Mean imprecision per ensemble size reached a plateau beyond 3-4 models. The best ensembles depended on the set of observations, comprising three models with the occasion 5 trough concentration and seven or eight models with occasions 1 to 4 observations. Across all sets, the ensemble optimised on the occasion 5 trough concentration was among the most precise, and score-based ensembles remained consistently precise, whereas the ensemble optimised on occasions 1 to 4 generalised poorly. Only a limited sampling strategy at occasion 5 kept imprecision below 20% for all ensembles. The plateau effect generalised to infliximab and cefepime. Ensemble composition matters more than ensemble size. Model averaging should rely on well-chosen ensembles rather than all available models.

Abstract Polycystic ovarian syndrome and Type 2 diabetes mellitus are complex multivariate diseases that share well-established clinical pathophysiology, yet the molecular basis of their overlap is unknown. Prior bioinformatics studies have identified common differentially expressed genes between these two conditions. However, they have not accounted for the directionality of the gene expression changes observed, potentially obscuring any biological distinctions they may have. This study applied a direction-aware approach to classify molecular signatures between PCOS and T2DM as concordant or discordant. Transcriptomic analysis of GEO datasets GSE138518 (ovarian granulosa tissue, PCOS) and GSE25724 (pancreatic islet tissue, T2DM) identified 225 and 1,302 DEGs, respectively. Venn diagram analysis showed that only three genes (SLC6A8, RGS4, and SORL1) were shared between PCOS and T2DM from the total of 1,527 genes, and all 3 genes were regulated in opposing directions. Disease gene retrieval from the Comparative Toxicogenomics Database, Online Mendelian Inheritance in Man database, and GeneCards showed 214 shared disease-associated genes, with 311 genes unique to PCOS and 496 genes unique to T2DM. Protein-protein interaction construction using STRING (version 12.0) identified 70 interacting nodes. CytoHubba analysis across 6 scoring methods identified 18 high-confidence hub genes, including INS, BCL2, MTOR, LEP, MFN2, and PIK3CD. Functional enrichment analysis identified biological processes including immune activation, apoptosis, and insulin signaling, which were confirmed as the main pathways in KEGG pathway analysis. These findings show that while PCOS and T2DM share limited DEGs, they share common pathogenic pathways.

Longitudinal non-Gaussian data are common in biomedical and microbiome studies, where responses may not be normally distributed. However, many existing dynamic graphical models are developed for Gaussian responses, while many non-Gaussian graphical models focus on cross-sectional dependence and do not explicitly estimate temporal transition effects. We propose a latent Gaussian dynamic graphical model for longitudinal non-Gaussian responses, including binary, count, and compositional outcomes. The proposed framework jointly estimates the contemporaneous conditional dependence network and the lag-one temporal dependence structure. We further develop a penalized Monte Carlo expectation maximization procedure for estimation. Simulation studies across multiple graph structures and outcome types show that the proposed method can recover both dependence structures, with more stable recovery for the lag-one temporal dependence structure and more structure-dependent recovery for the contemporaneous conditional dependence network. We apply the proposed method to longitudinal data from the Integrative Human Microbiome Project (HMP2) using a multinomial observation model. The results reveal persistent lag-one temporal dependence and contemporaneous conditional dependence among dominant microbial taxa, demonstrating the utility of the proposed framework for dynamic network analysis of longitudinal non-Gaussian biomedical data.

Although largely preventable, cardiovascular disease (CVD) is the leading cause of death worldwide. Synthetic data generation (SDG) of health data can facilitate data sharing for research and development of personalised cardiovascular disease prevention tools. Current state-of-the-art SDG methods are limited to cross-sectional data or focused on short-term or in-patient EHR applications lacking benchmarks for generalisability. Since prevention is a long-term longitudinal objective, these limitations must be addressed. We evaluated two tabular generative adversarial networks, TGAN and CTGAN, for privacy and fidelity on a structured 6-year longitudinal trial dataset within CVD prevention. We compared performance with a rudimentary perturbation method. In addition, we performed a variable-wise evaluation and an age-stratified analysis of the higher performing model to assess preservation of clinically relevant characteristics. Overall, CTGAN showed better fidelity than TGAN. The models performed similarly within privacy, showing better results than the perturbation method. CTGAN had lower performance in comparison to the perturbation method within numerical fidelity of continuous variables and overall temporal fidelity. Furthermore, CTGAN's performance was lower for younger age groups than older. CTGAN shows better than or equal performance within fidelity and privacy than TGAN and the perturbation method on a long-term longitudinal dataset within CVD prevention. To the best of our knowledge, this is the first study evaluating CTGAN on a longitudinal dataset and our findings suggest that handling of temporal characteristics and imbalanced subgroups need further refinement.

Several tools are available to detect differential gene expression. The predominant approach of different statistical methods is to consider changes in gene expression independent from their genomic context. Previously, DELocal was introduced that takes into consideration gene expression dynamics in the genomic neighbourhood. Whereas DELocal was shown to identify developmentally regulated genes in developing mouse tooth, its performance in other biological conditions remains undetermined. To benchmark DElocal together with other methods, we develop another tool 'broadSeq'. It provides easy to use interface for methods like DELocal , deseq2, limma, edger, EBSeq and many more. Here we report benchmarking results on different data sets that comprise wild type developing mouse organs from different embryonic time points, and also human cancer data. The examined datasets are from the heart, kidney, liver, brain, forebrain, and hindbrain. We found that most methods produce relatively comparable lists of differentially expressed genes. Nevertheless, DELocal identifies partially distinct sets of genes compared to the other methods. Evaluating the identified genes using gene and disease ontology terms show that the uniquely detected genes by DELocal are biologically relevant and also highly expressed compared to the genes identified by the other methods. Because consistently highly expressed genes show relatively subtle fold changes, the neighbourhood concept of DELocal appears to correct for the difficulty in detecting these differential expression of highly expressed genes. Finally, we provide the tools as combined bioconductor packages, allowing researchers to explore and decide the best methods for their data and questions.

An Ischemic stroke is one of the main causes of death and long-term disability around the world. There are very few treatment options available, especially for the cases where the short time period to give clot dissolving medicine has passed (Powers et al., 2019). While single-cell RNA sequencing (scRNA-seq) has emerged as a powerful tool for understanding cell-to-cell differences within at issue (cellular heterogeneity) (Tang et al., 2009). However translating these high dimensional datasets into clinically useful biomarkers remains challenging (Rajko-mar et al., 2019). A major problem with current biomarker studies is their reliance on a single machine learning model, where results may reflect model bias rather than true biological signals (Lopez-Rincon et al., 2020). Methods: We developed a multi-model consensus framework integrating eight different machine learning methods from three groups: traditional models (logistic regression, random forest, XGBoost), a hybrid deep learning model (CNN-XGBoost), and transformer-inspired models (scGPT, scBERT, sc-Former, Geneformer). The proposed framework was applied to scRNA-seq data from a mouse middle cerebral artery occlusion (MCAO) model (GSE174574; n=6 mice, 54,599 cells). To prevent data leakage, we implemented rigorous subject-wise train/test splitting, ensuring all cells from individual animals remained either in the training sets or in the testsets. Results: All eight models achieved excellent performance on the independent test data, with AUC scores between 0.9778 and 0.9975. From six models that allowed gene level importance extraction, we identified 63 consensus genes (threshold: [≥] 4/6 models), with 17 genes selected by all 6 models. Pathway enrichment analysis showed significant enrichment of interferon signaling, with a novel dual-interferon pattern. Twelve of the consensus genes respond to interferon, and five of them (Ifitm6, Ifitm1, Gbp2, Igtp, Tgtp1) respond to both Type I (IFN-/{beta}) and Type II (IFN-{gamma}) interferons. Notably the consensus included Il1rn (interleukin-1 receptor antagonist), the target of Anakinra -- an FDA-approved drug that has completed Phase II stroke trials with promising safety profiles (Smith et al., 2018; Emsley et al., 2005). Conclusions: The Multi-model consensus framework, proposed in this paper, provides a robust and reliable approach to biomarker discovery that goes beyond the limitations of any single method and reduces individual algorithm bias. The convergence of classical, ensemble, and deep learning methods on interferon and IL-1 pathway genes supports their biological validity as stroke biomarkers. Our findings provide a prioritized gene list for further clinical validation and highlight interferon signaling as a possible treatment strategy warranting further investigation. Keywords: ischemic stroke; machine learning; single-cell RNA sequencing; biomarker discovery; consensus analysis; interferon signaling; neuroinflammation

In 1998 we predicted percent human intestinal absorption (HIA) for 86 drug and drug-like compounds from six molecular descriptors chosen by a genetic algorithm with a small neural network serving as the fitness function. That result rested on one division of the compounds into three sets: training, cross-validation and external prediction. Repeating the division shows it was a rather fortunate one. An extraction is one random draw from a wide distribution, so we repeat every experiment 250 times and report the result as a distribution. Across 250 stratified extractions, 14 match the published cross-validation error and only one matches the published external prediction error. Reconstructed this way, the published six descriptors give a mean extraction error roughly twice the external prediction value we reported in 1998 (Table 1). We then applied the same six descriptors to a larger set of 780 compounds from a public compilation. Under the 1998 network they give 22.50 %HIA units on training compounds against 25.77 on extracted compounds, whereas on the 86 they give 13.43 against 30.39. More compounds close the gap between fitting and prediction. Four datasets are modeled here: 780 small molecules with HIA measured in man, 3,771 small molecules and 1,281 cyclic peptides measured across Caco-2 cell monolayers, and 7,298 cyclic peptides measured across artificial membranes. Each is its own prediction problem, and one procedure is applied to all four. A genetic algorithm chooses which descriptors to use and how many, and a random forest grades each choice on compounds its own trees never saw. The descriptors chosen most often across the 250 extractions form one consensus set for each dataset. For absorption, 25 consensus descriptors under a random forest reach an extraction rms of 20.47 %HIA units at R 0.76. On the same compounds and the same extractions, the six 1998 descriptors under the 1998 network reach 25.77 at R 0.59. The four models are served at predicthia.ai, where a structure can be drawn or pasted, scored against its training set for similarity, and the training data with fitted models downloaded. Each model served is its fixed consensus set fitted on every compound in its dataset; the accuracy quoted beside a prediction comes from evaluating that same set across 250 extractions (Table 3). The fixed sets hold 25 descriptors for absorption, 55 for Caco-2 small-molecule permeability, 50 for Caco-2 peptide permeability and 25 for artificial-membrane permeability. Keywords: human intestinal absorption; QSPR; genetic algorithm; descriptor selection; cyclic peptides; membrane permeability; oral bioavailability; repeated extraction

Rapid and accurate identification of viral sequences underpins clinical diagnostics, epidemiological surveillance, and the safety testing of biological products, yet the established alignment-based methods such as BLAST are inherently closed-set, with accuracy tied to how well a query is already represented in the reference database. Recent advances in genomic foundation models (GFMs) have enabled alignment-free approaches to sequence classification. However, their performance relative to traditional methods remains incompletely characterised. We evaluated GFMs, specifically the Nucleotide Transformer (NT) and DNABERT families, for viral sequence classification using a dataset derived from the Reference Viral Database (RVDB). Sequence embeddings were generated with mean, max, and CLS pooling, classified by nearest-neighbour search in an embedding index and compared with BLAST as a reference. Matching BLAST provided a particularly stringent benchmark here, because as a closed-set method it compares each query directly against the exact reference sequences it stores, whereas a foundation model must rely on representations shaped by broad, general-purpose pre-training. On sequences of standard length, the best model, NTv3-650M, reached an accuracy of 0.93, approaching the 0.97 achieved by BLAST. On long genomic sequences, evaluated on a much smaller test set, the difference narrowed further, with NTv3-650M reaching 0.95 against 0.98 for BLAST. Robustness analyses showed that all GFMs were markedly more sensitive to base masking than BLAST. Both methods indexed the same labelled reference sequences, and BLAST compared queries with them directly at the nucleotide level, so approaching it from frozen, general-purpose representations without any parameter update is a meaningful result. NTv3-650M emerged as the most promising candidate for further optimisation and deployment in bioinformatics applications.

Background Multi-omics integration has emerged as a powerful strategy to decode the molecular complexity of biological systems. However, the diversity of available methods, each designed with distinct assumptions, objectives, and computational requirements, makes method selection, usage and interpretation challenging for nonexpert users. Here we present BiomiX 3.0, an updated version of the BiomiX platform that extends its integration capabilities with four additional methods: Similarity Network Fusion (SNF), NEighborhood-based Multi-Omics clustering (NEMO), Data Integration Analysis for Biomarker discovery using Latent variable approaches for Omics studies (DIABLO), and PRAMIGO (Phenotyping netwoRk Application for Multi-omics InteGratiOn), a novel supervised heterogeneous graph transformer (HGT) introduced in this work. Results We benchmarked all five methods, MOFA, DIABLO, SNF, NEMO, and PRAMIGO, on two independent multiomics datasets derived from a Chronic Lymphocytic Leukemia (CLL) cohort comparing IGHV-mutated and unmutated patients, and a pulmonary tuberculosis (PTB) cohort versus healthy controls. Supervised methods (DIABLO, PRAMIGO) consistently achieved higher condition-specific discrimination as measured by the Adjusted Rand Index (ARI) and the Adjusted Mutual Information (AMI). In contrast, unsupervised methods (SNF, NEMO) revealed alternative patient stratifications driven by independent sources of biological variance while MOFA performed in a semi-supervised way occupies an intermediate position, capturing latent factors that explain both disease-associated and orthogonal sources of variance. Systematic gene-centric analysis of the top-ranked features prioritized by each method was supported by manual biological annotation of shared and method-specific signals. Across cohorts, we annotated 154 shared features (116 genes, 38 metabolites) and 60 method-unique features per cohort, demonstrating that no single integration strategy captures the full landscape of biologically relevant signals. In the CLL cohort, shared features spanned B-cell receptor biology, innate immune signaling, RAS/MAPK activation, and epigenetic regulation, while methodunique features revealed supervised-method-specific insights into vesicle trafficking (DIABLO), immune checkpoints (MOFA), and ncRNA regulation (PRAMIGO). In the PTB cohort, a convergent interferon/innate immune signature dominated shared features across all methods. Still, method-unique analysis uncovered DIABLO-specific acylcarnitine metabolic reprogramming, MOFA-specific restoration of lysosomal trafficking, and PRAMIGO-specific {gamma}{delta} T-cell and immunoglobulin repertoire diversity. Across both cohorts, SNF and NEMO proved useful for detecting biological and technical sources of variation that were orthogonal to the primary condition of interest. Ultimately, PRAMIGO uniquely enables the construction of heterogeneous graphs modeling cross-modal molecular interactions, uncovering epigenetic co-regulation programs in CLL and multi-omics inflammatory modules in PTB that are difficult to identify using conventional integration approaches. Conclusions BiomiX 3.0 provides a graphical user interface (GUI) multi-method integration environment that democratizes access to state-of-the-art multi-omics analysis. By combining both unsupervised and supervised integration strategies within a unified platform and introducing graph-based learning through PRAMIGO, BiomiX 3.0 enables researchers across disciplines with complementary tools to interrogate the biological sources of variation in their data, without requiring bioinformatics expertise.

In the recent growing interest of AI research toward biology, genetic association studies of AI- derived phenotypes from high-content modalities such as images emerges as a powerful means for biological discovery. However, such AI-phenotyping methods still lacks a good optimization target and an efficient evaluation framework. The number of discovered loci was the major criterion for evaluating the quality of AI-derived phenotypes. However, the time and computational resources required for running the GWAS and subsequent loci-clumping are substantial, limiting the rapid development and iteration of deep learning representation algorithms. Here we present a 1000x faster and lightweight framework, sum-h2, than traditional GWAS framework for evaluating genetic discovery through total heritability. We revisit tr(P^(-1) G), previously proposed in the context of evolutionary studies, as a measure of multi-phenotype heritability. We showed that sum-h2, the sum of heritability over phenotypic Principal Components (PCs), is equivalent to the linear transformation-invariant tr(P^(-1) G), through both theoretical proof and simulation studies. Moreover, sum-h2 can be estimated rapidly with minimal information loss over a relatedness-enriched sample, while preserving the relative ranking of endophenotypes by GWAS loci counts. Based on selected UKB data and sum-h2, we set up a Genetic Discovery Arena, enabling rapid and fair comparisons for the development and optimization of deep learning-derived phenotyping methods.