BMC Bioinformatics

Background. Patient stratification from multi-omics and clinical data is essential for uncovering disease heterogeneity and moving toward more personalized treatment strategies. However, integrating heterogeneous data layers while identifying robust patient strata remains challenging. Methods. We introduce Reduced Fusion of Multi-Omics Stratification (RedFuMOS), a novel three-step approach for patient stratification based on mixed-type multi-omics data. RedFuMOS extends Similarity Network Fusion to accommodate mixed-type data layers and layer-specific similarity measures for data integration, includes a dimensionality reduction step to mitigate the curse of dimensionality, and performs patient stratification using density-based hierarchical clustering with HDBSCAN. It also implemented an automated optimization procedure to identify the best set of hyperparameters, minimizing the need for manual tuning. Results. RedFuMOS outperformed six state-of-the-art tools for multi-omics patient stratification in a comprehensive simulated benchmarking study, which also confirmed that, although computationally expensive, the dimensionality reduction step is crucial for achieving good stratification performance. Additionally, RedFuMOS identified two clinically relevant patient strata in a small real-world cohort of patients with Philadelphia chromosome-positive chronic myeloid leukaemia. Conclusion. RedFuMOS provides a flexible framework for integrating heterogeneous multi-omics and clinical data. RedFuMOS is available as an R package at http://github.com/delucasara/RedFuMOS.

Network-based analyses of molecular interactions are useful for interpreting high-throughput omics data and identifying therapeutic targets. Cytoscape is the standard platform for these tasks, but users face a trade-off between accessible graphical workflows that are difficult to document and reproducible automation in Python or R that requires programming expertise. General-purpose coding assistants can generate Cytoscape Automation scripts, but remain external to Cytoscape. We present CyChat, a Cytoscape Desktop app that integrates a chat interface and a large language model (LLM) agent into the application. CyChat translates natural language into executable Cytoscape Automation workflows, runs generated Python code, and exports chat sessions with executed code as standalone Jupyter notebooks. To reduce setup barriers, CyChat includes an embedded Python runtime and supports both cloud-based and locally hosted LLMs. CyChat was evaluated across ten Cytoscape workflows using seven LLM providers, each represented by one LLM. The strongest configuration achieves a pass rate above 99%. In a qualitative evaluation based on a published network visualization, CyChat completes the task in 1.5-5 minutes, compared with 15-20 minutes for manual GUI workflows by computational biologists. CyChat is available through the Cytoscape App Store at https://apps.cytoscape.org/apps/cychat.

Three-dimensional (3D) mural cell morphology is heterogeneous and coupled to vessel geometry, however, measurements from two-dimensional (2D) maximum intensity projections (MIP) obscure overlapping processes and cell-vessel contacts. Accordingly, we developed Mural-VISTA, a semi-automated Python workflow for mural cell-vessel interaction and single-cell topo-morphology analysis of reconstructed surface meshes. This workflow integrates mesh pretreatment, interactive centerline extraction, hierarchical segmentation of cell soma, main axis and secondary processes (branches), and extraction of 36 multiscale (cell process segment level, process level, and whole cell level) topo-morphological and vessel-referenced metrics. Mural-VISTA identified morphological changes in pericytes and vascular smooth muscle cells (vSMCs) with altered RhoA activity. Constitutive active RhoA (RhoA CA) over-expression reduced branch complexity and increased process alignment in both cell types, while increased whole-cell and branch solidity only in vSMCs. Dominant negative RhoA (RhoA DN) over-expression increased branch abundance and reduced branch solidity in pericytes but not vSMCs, suggesting cell-type specific effect of reduced RhoA activity. In conclusion, Mural-VISTA enables quantitative 3D profiling of mural cell architecture and its spatial relationship with the vessel.

Amplicon sequencing of the 16S rRNA gene is the most widely used approach for profiling bacterial communities, but its taxonomic resolution is typically limited to the genus level. Many species carry multiple divergent 16S rRNA alleles that overlap across species boundaries, an ambiguity that even full-length, long-read sequencing cannot fully resolve. Shotgun metagenomics achieves species-level resolution but remains costly, particularly when only a single genus is of interest. Amplicon sequencing of rapidly evolving, protein-coding housekeeping genes offers a cost-effective alternative, yet no tool exists to identify suitable primer sets for a given target taxon. Here we present AmPair, a Snakemake pipeline that, given a target genus and one or more candidate housekeeping genes, designs and ranks primer pairs binding conserved regions while flanking a variable region capable of species-level discrimination, and validates them in silico across all available genomes. Using the genus Bacillus and the housekeeping gene tuf as a case study, the primer set recommended by AmPair amplified 99% of 2,392 genomes; only 0.04% carried multiple alleles and none showed inter-species allele overlap, compared with 91.41% and 69.49%, respectively, for the standard 16S rRNA V1-V9 region. Applied to a Bacillus community profiled by Nanopore sequencing, the same primers resolved closely related species. AmPair thus offers a generalizable and accessible route to species-level community profiling.

Causal mediation analysis is widely used to identify biological pathways linking exposures to outcomes, but most methods assume homogeneous mediation effects across individuals. In high-dimensional omics settings, this assumption can mask important heterogeneity driven by demographic, genetic, or environmental factors. We propose the M-high-learner, a flexible framework for detecting heterogeneous mediation effects with high-dimensional mediators. The method identifies mediators with subgroup-specific indirect effects while distinguishing them from null or homogeneous signals and controlling the type I error rate. It is computationally efficient, scalable, and yields interpretable sub-types. Simulation studies show that the proposed approach achieves high power while maintaining accurate error control. Applications to the Framingham Heart Study and the Multi-Ethnic Study of Atherosclerosis reveal that the mediation role of gene expression in sexs effect on high-density lipoprotein varies across subgroups defined by body mass index and age. Our framework provides a practical tool for uncovering heterogeneous biological mechanisms in high-dimensional genomic studies. Author SummaryBiological processes linking risk factors to disease often differ across individuals, but many existing methods assume these processes are the same for everyone. This can hide important differences between groups. We developed a powerful method to identify when these pathways vary across subgroups using large-scale molecular data. Our approach detects differences in how intermediate biological factors contribute to outcomes in populations defined by characteristics such as age and body mass index. Applying our method to population studies, we found that some biological pathways operate differently across groups, suggesting that key mechanisms may be missed when differences are ignored. Our work provides a tool to better understand how disease-related processes vary across individuals, which may support more targeted and personalized approaches to health research.

Bioinformatics software and databases are essential components of modern life science research, yet their mentions in the scientific literature are often inconsistent and difficult to systematically identify at scale. The lack of a comprehensive and up-to-date catalog of bioinformatics resources hinders efforts toward automated biomedical knowledge extraction and streamlined data analysis. Here we present SNAIL, a hybrid named entity recognition framework designed to automatically identify bioinformatics software and database (SW/DB) names from biomedical texts. SNAIL integrates complementary lexical and semantic modeling strategies. The lexical component captures orthographic patterns and contextual cues characteristic of SW/DB names, while the semantic component leverages contextual embeddings generated by transformer-based language models such as SciBERT, combined with an explicit token-masking strategy to enhance entity-focused representations. A large training corpus was constructed automatically through a hybrid pipeline that integrates citation-hinted extraction with large language model-assisted distillation. Evaluation on two independent benchmark datasets and real-world research articles demonstrates that SNAIL substantially outperforms existing approaches, including domain-specific methods such as bioNerDS2 and general-purpose large language models such as ChatGPT, Gemini, Grok and Claude. Applying SNAIL to large-scale literature analysis further reveals distinct journal-level preferences across bioinformatics subfields. These results demonstrate that SNAIL provides an accurate and scalable solution for identifying bioinformatics resources in scientific texts and enables systematic meta-analysis of tool usage and research trends.

Background: Reproducible taxonomic collapsing and geological-timescale annotation of time-calibrated phylogenetic trees in R often require coordination among several packages and repeated code for label parsing, clade validation, plotting, and export. Workflow-managed analyses additionally benefit from non-interactive configuration, predictable diagnostics, and machine-readable exit status. Results: We present Rclade, an R package that consolidates the multi-package coordination required for taxonomic collapsing into a streamlined, single-function interface. Rclade provides (1) custom ggproto objects (GeomPolygonStraight/GeomSegmentStraight) that bypass coord_munch() interpolation to achieve straight-edge rendering of collapsed triangles in circular layouts; (2) automatic detection and parsing of four taxonomic-label formats (GTDB, Silva, NCBI, embedded) plus user-supplied custom regex, with explicit input-validation contracts and parsing-accuracy evaluation on real and derived test sets; and (3) workflow embeddability through YAML configuration, library-mode APIs, and standard Unix exit codes. Benchmarks on synthetic and real datasets (200-10,000 synthetic tips and real reference trees up to 10,122 tips; 5 replicates at every scale under a unified fully rendered measurement protocol) show that the full-pipeline overhead is modest for interactive use (median {approx}0.87 s in-session rendering and {approx}8.4 s process-level wall-clock at 10,000 tips). Conclusions: Rclade is a convenience layer over the ggtree/deeptime ecosystem that reduces boilerplate while adding targeted technical improvements for circular-layout rendering and format heterogeneity management.

Scientific agents can produce plausible answers while remaining unable to establish whether the computation behind an answer is executable, recoverable, or reproducible. We present BloClaw, an AI4S workstation built around a simple principle: a scientific agent should know what it can do, show how it did it, and state what remains unvalidated. Each capability declares an execution state, input constraints, dependencies, expected outputs, and scientific limitations. Natural-language requests are translated into structured tasks, validated against this registry, executed through scientific tools, and recorded in a provenance-aware Living Lab Notebook. The system is designed to detect invalid inputs, failed tool calls, missing dependencies, and remote timeouts, and to route them to repair, retry, or escalation. The implemented and tested scope comprises RDKit-based molecular property and rule screening, protein structure analysis, docking-pose inspection, 3D visualization, and structured reporting. We demonstrate the workflow on a PubChem-retrieved osimertinib structure and a supplied 6LU7 docking artifact: the former yields deterministic descriptors (molecular weight 499.619 Da, cLogP 4.5098, TPSA 87.55 A^2), while the latter contains 2,387 protein ATOM records, 309 residues, and nine pose records. These examples are workflow demonstrations, not efficacy or affinity studies. Beyond retrospective prediction, the manuscript specifies a prior-minimized constructive mode in which a desired function is compiled into explicit physical, chemical, and systems constraints, candidate mechanisms are simulated, and observations are reintroduced for calibration and falsification; this is a proposed extension rather than a result of the present case studies. We describe an evaluation protocol that compares BloClaw with a standard single-agent workflow and fixed-script execution using task completion, scientific correctness, recovery success, provenance completeness, reproducibility, human review time, latency, and cost. This manuscript reports the system design, verified capability boundary, deterministic software artifacts, and a reproducible evaluation protocol; it does not claim benchmark improvements before those experiments are run. BloClaw is an execution and accountability layer for AI-assisted research, complementing expert review and experimental validation rather than replacing them.

We describe LRSPAT (low-rank spatial toolkit), a fast and memory-efficient framework for approximating measures of spatial association for high-dimensional data. While LRSPAT can be applied to any multivariate spatial dataset, development was motivated by the computational challenge of identifying spatially variable genes in high-resolution spatial transcriptomics (ST) data generated by technologies such as 10x Visium HD, Xenium and Atera. LRSPAT leverages a truncated SVD of the expression data and a thresholded spatial weights matrix to perform reduced-rank reconstruction of spatial statistics in the quadratic form family, including global and local versions of Moran's I, Geary's C, and Getis-Ord G. A regularization approach is leveraged to account for the inflated null distribution of spatial statistics computed on latent variables. By performing key operations on the low-dimensional embeddings, LRSPAT is orders of magnitude faster than standard implementations with significantly lower memory requirements. Because the low-rank approach denoises and desparsifies ST data, LRSPAT is also more accurate than standard techniques at identifying genes with true spatial expression patterns. The dramatic improvements in execution time and memory consumption enable the genome-wide analysis of spatially variable genes (SVGs) and exploration of the full range of hyperparameters including spatial scale, distance metric, and embedding rank. This preprint outlines the background and mathematical details of the approach with limited preliminary results and a short conclusion.

Motivation: A continued decrease in sequencing costs has facilitated the exponential increase in available sequencing data, with public databases like the European Nucleotide Archive (ENA) and Sequence Read Archive (SRA) reaching well in the order of petabases. This has been the incentive to develop more scalable tools for common bioinformatics tasks. One such task is the approximate searching of short sequence patterns like genes or reads in reference data sets. In recent years, a variety of indexing data structures have been proposed for searching large sequencing databases. The state-of-the-art index, the Hierarchical Interleaved Bloom Filter (HIBF) was first-in-class to index one million samples. To be useful for expanding repositories, it must be extended to support dynamic updates. Results: In this paper, we introduce a scalable and updatable sequence-search index by extending the HIBF with partial rebuilding to support efficient updates. We demonstrate the Dynamic HIBF's capacity for large-scale data by iteratively creating an index from over 100 TB of compressed reads across more than 39,000 full human RNA-Seq samples, updated in consecutive batches of 100. To benchmark against state-of-the-art tools, we evaluated incremental performance on a subset of 5,000 samples sub-sampled to 1% of their original read depth. In this comparative setting, the dynamic HIBF completed the sequential insertion of all 5,000 samples within 5 hours--24 to 65 times faster than competing methods and twice as fast as the static HIBF.

Constructing microbial association networks is a common strategy for exploring relationships among taxa in microbiome studies. Although marginal correlation methods are easy to implement and allow formal inference, they can produce spurious edges driven by indirect associations through other taxa. Conditional graphical-modeling methods aim to recover direct associations, but many rely on Gaussian or linear assumptions and often provide limited uncertainty quantification. We propose a conditional, nonparametric approach based on the scaled expected conditional covariance (SEcov). SEcov measures population-level conditional association by residualizing each taxon with respect to the remaining taxa and scaling the resulting expected conditional covariance. The resulting estimator can incorporate flexible machine-learning methods for conditional-mean estimation and admits asymptotic normal inference, enabling p-values and confidence intervals for taxon-pair associations. We demonstrate through simulation studies that our proposed approach improves network recovery relative to other methods, and we illustrate the new method via construction of a co-occurrence network for the vaginal microbiome during pregnancy. IMPORTANCEHigh-throughput sequencing has made it possible to characterize microbial communities at large scale, and network analysis is widely used to summarize relationships among taxa. However, networks based on marginal correlations may include indirect associations, whereas many conditional graphical models rely on assumptions that may be difficult to justify for sparse, zero-inflated, compositional microbiome data. SEcov offers a practical alternative by estimating conditional associations nonparametrically and attaching inferential uncertainty to individual edges. This allows investigators to construct microbiome networks using statistically interpretable evidence for taxon-pair associations, rather than relying solely on arbitrary correlation cutoffs or regularization tuning parameters.

The dynamics of gene regulatory networks are governed by intrinsic noise, stemming from the random nature of biochemical reactions, and by extrinsic noise, arising from fluctuations in cellular components and environmental conditions. Together, these sources can compromise the reliability of predictive computational models if not properly accounted for, and capturing both effects within a single framework remains a non-trivial task in computational biology. In this work, we propose an uncertainty quantification framework that addresses these two contributions jointly: intrinsic stochasticity is described through a partial integro-differential equation (PIDE) for the protein probability density function, whereas extrinsic noise is represented as parametric uncertainty in the kinetic parameters. The propagation of the uncertainty is carried out via an intrusive polynomial chaos expansion (PCE), in which the PCE coefficients are obtained from a stochastic Galerkin projection of the PIDE, yielding a coupled deterministic system that is solved with standard numerical methods. We illustrate the approach on a positive autoregulatory gene network with one and two uncertain kinetic parameters. The proposed approach accurately reproduces the mean, variance, and full protein probability density function, including the bimodal distributions, at a substantially lower computational cost.

Motivation: Interactive tools that let non-programmers explore an analyzed single-cell dataset, its embeddings, gene expression, cell metadata, and marker genes, have become standard laboratory infrastructure. Every actively maintained tool in this space (ShinyCell, ScRDAVis, sCIRCLE, scViewer) is built on R Shiny and requires a Seurat object as input. Laboratories whose primary analysis pipeline is Python/scanpy, the dominant framework for single-cell RNA-seq, spatial, and multi-omic analysis, therefore have no lightweight, language-native option that pairs a shareable web-based viewer with a scriptable Python API: sharing a scanpy result means either exporting to Seurat first or handing over a notebook that only a programmer can run. Results: We present scPyviewer, a web-based viewer that ingests AnnData objects directly and reproduces the core interaction patterns of the incumbent R Shiny tools without leaving the Python stack. In a feature-parity audit against three actively maintained R Shiny incumbents, scPyviewer matches or exceeds every baseline capability (7/7); among these, it uniquely offers native AnnData ingestion with no Seurat conversion, and cross-dataset comparison over shared genes and matched cell-type composition. Benchmarked head-to-head against the R/Seurat rendering substrate the incumbents are built on, identical operations, identical data, across three datasets spanning 22,315 to roughly 313,000 cells, scPyviewer renders every core view faster at every scale tested (up to 3.6x on a single view) and at a fraction of the memory (5.2x lower on the smallest dataset). At the largest scale tested, the gap becomes categorical rather than incremental: scPyviewer completes every view on a 313,000-cell dataset while the Seurat substrate exhausts an 8 GB memory budget and fails outright. Beyond the interactive app, scPyviewer installs via pip or conda and exposes a public Python API that returns Matplotlib figures and pandas tables for scripted, publication-ready output. Availability and implementation: scPyviewer is implemented in Python 3.11 (scanpy 1.11.5, anndata 0.12.19, streamlit 1.59.2, plotly 6.9.0) and distributed with a one-command reproduction interface that installs pinned dependencies, regenerates the benchmark and all figures, and launches the interactive app. Source code is available at https://github.com/xuan13hao/scPyviewer.git.

Mechanistic mathematical models have been used extensively to provide a deeper understanding of biological mechanisms, including unveiling the regulation of tumour growth and its response to various treatments. However, given the breadth of biological regulatory mechanisms, these models are frequently large and thus prone to potential issues with parameter identifiability. Statistical metrics like the Akaike and Bayesian information criteria can help identify a parsimonious model by balancing goodness of fit against model complexity. Yet simple models may fail to provide sufficient biological insight if they do not adequately capture known physiological processes or mechanisms. A modeller must therefore balance hypothesis generation and biological learning with model tractability. Here, we illustrate this balance using models of ovarian cancer growth and treatment response to cisplatin and immune checkpoint blockade in homologous recombination (HR)-deficient and HR-proficient immunocompetent mouse models. We develop a hierarchy of mathematical models of increasing complexity to describe tumour growth, treatment response, and immune dynamics. Our results highlight the limits of relying purely on statistical metrics for model selection, particularly when the goal is to obtain biological insight and underscore the importance of balancing model complexity to avoid overfitting and parameter unidentifiability.

Identifying causal connections between existing drugs and mechanistic profiles of diseases is a foundational step for effective drug repurposing. Although knowledge graphs (KGs) are highly suited for consolidating biomedical databases and tracking these connections, a single biomedical KG is constrained by its ingestion pipeline and knowledge sources. While different biomedical KGs could be complementary if combined, efforts to combine them into a unified and more comprehensive KG are hindered by lack of interoperability and poor provenance. To address those issues, we present EC-KG, a Biolink Model-compatible KG for computational drug repurposing. EC-KG is an interoperable, provenance-first KG which integrates RTX-KG2, ROBOKOP, and PrimeKG at the network-level, encapsulating over 7 million nodes and 81 million edges from 95 primary data sources. EC-KG has improved coverage of core biomedical entities such as drugs, targets, and diseases relevant to drug repurposing vs source graphs, and captures complex biomedical mechanisms within its topology. We demonstrate that the network unification in EC-KG leads to emergence of novel, mechanistically relevant pathways which are disconnected in the underlying constituent networks and show its applications in method development, benchmarking and predictive drug repurposing applications. EC-KG has already been successfully used in drug repurposing research to surface Botulinum Toxin A as a candidate to treat Major Depressive Disorder, as well as to validate repurposing of Lenalidomide and Dexamethasone for a subgroup of patients with Rosai-Dorfman Disease.

Genome-wide coexpression analysis of intact tissue samples is a powerful approach for identifying reproducible signatures of cell types and states, since it can survey vast numbers of individuals, cells, and transcripts. However, it can be difficult to optimize gene coexpression network construction and compare results from independent analyses. To address these challenges, we developed OMICON (theomicon.ucsf.edu) for research on human brain gene coexpression networks. OMICON contains gene expression data from >17K normal and neoplastic human brain samples with standardized metadata. Systematic analysis of independent datasets identified >250K gene coexpression modules, which were characterized and compared via enrichment analysis with >40K gene sets. All modules are discoverable via an advanced search engine that can filter by genes, metadata, and enrichment results. Analyses can also be browsed with an interactive workflow visualization tool, and users can communicate within OMICON using @mention functionality to support communal research on human brain gene coexpression networks.

The interpretation of missense variants remains a major challenge in clinical genetics. "Meta-domains" aggregate population and pathogenic variation across homologous Pfam domain instances in the human proteome, providing per-residue context for interpreting variants of uncertain significance (VUS). Our 2019 implementation, MetaDome, is widely used and named in clinical variant-classification guidelines. Here we present the MetaDome 2027 update, featuring a comprehensively updated dataset and GRCh38 support. The redesigned pipeline enables incremental updates of GENCODE, UniProtKB/Swiss-Prot, Pfam, gnomAD, and ClinVar while maintaining 100% sequence-identity gene-to-protein mapping. Annotated Pfam domain instances grew 14.9% from 71,419 to 82,069 and meta-domain-eligible Pfam families ([≥]2 human occurrences) by 73.3% from 3,334 to 5,778; Pfam domains are annotated to 92% of human proteins. Approximately 43% of mapped protein-coding nucleotides (14.3 million in GRCh38, 13.8 million in GRCh37) are in a meta-domain; in GRCh38 67.9% (37,692 of 55,548) of pathogenic or likely pathogenic ClinVar missense variants fall at such a position. We show how MetaDome helped reclassify a de novo missense VUS in RALA and identify 52,463 ClinVar missense VUS for which meta-domains supply otherwise unavailable pathogenic evidence. MetaDome is freely available at www.metadome.app.

Thermodynamic models quantitatively describe interactions between transcription machinery and bacterial promoters. Contrary to conventional understanding, model analysis by Parisutham et al. (2025) attributes transcriptional inhibition by repressors to overstabilization of the RNA polymerase-promoter complex rather than prevention of its formation. Moreover, it suggests an inverse scaling relationship between basal promoter strength and transcriptional fold change, applicable to both repressor- and activator-mediated regulation. To reevaluate findings from this study, we systematically analyze empirical data and compare its framework with conventional thermodynamic models. In contrast to the inverse scaling relationship, data across multiple sources exhibit a peaked tradeoff between basal promoter strength and fold change, underscoring the importance of broad data coverage in revealing the full pattern required for reliable model inference. Furthermore, we identify the model assumption responsible for the apparent inverse scaling and misinterpretation of regulatory mechanisms. Relaxing this assumption enables the model to capture the peaked tradeoff and yield inferences consistent with established mechanisms of transcriptional repression and activation. We further derive a mathematical solution that connects basal expression to fold change for both repressor- and activator-regulated promoters. Our results underscore the importance of broad data coverage to avoid a blind-men-and-elephant interpretation and establish basal promoter strength as a key design parameter governing transcriptional regulation.

Bacteriophages can only be understood through their interactions with bacterial hosts. As environmental sequencing efforts expanded, the number of available phage genome sequences has exploded, yet the vast majority of these sequences lack host information. Predicting the host of a newly observed phage is therefore a key challenge in virology. Several computational tools can predict phage-host relationships from genomic data, but they share notable limitations: (1) the number of different hosts that can be predicted remains relatively restricted; (2) tools tend to assign confident host predictions to non-viral input sequences; and (3) most tools have a trade-off between accuracy and speed. Here we present PhageTransformer (PT), a deep learning model for phage-host prediction that addresses these limitations. We benchmark PT against existing tools on 3,881 independent phage-host pairs from GenBank and public HiC data, and demonstrate that it achieves competitive or superior prediction accuracy at greatly reduced runtime.

Improvements in data availability, sharing, and integration, together with the development of explainable artificial intelligence (XAI) techniques, are advancing precision medicine for pediatric cancer by facilitating diagnosis, biomarker discovery, and drug development. Data sharing commons and initiatives like the Childhood Cancer Data Initiative (CCDI) provide access to pediatric-specific genomic and clinical data cohorts and improve data availability for pediatric cancer research. Based on CCDI, a scalable AI platform, Graph Artificial Intelligence for Pediatric Oncology (GAIPO), integrates various data modalities from bulk and single-cell omics data to clinical information. Such multi-modal data facilitates the training and development of advanced XAI models for pediatric cancers. We then developed an end-to-end multi-modality framework, PCGS, for pediatric cancer by incorporating omics-specific representation learning via GNN models with cross-attention fusion and multi-objective learning for downstream tasks such as classification, clustering, and survival analysis. This framework outperforms previous supervised multi-omics integration baseline approaches based on glioma and Wilms tumor cohorts and enables GNN model explainability via Shapley value-based feature attribution approaches to explain the contributions of gene-level features across various biomedical tasks, including classification and survival. Given specific background samples (e.g., age groups, sex, grades) as baselines, this explainable GNN model estimates and ranks the importance scores for input features from each omics modality. It identifies background-specific key features for biomarker discovery, risk group identification, and survival analysis in glioma and Wilms tumor, with potential applicability to other pediatric cancers.