Biology subjects

Ewing, R.

Publications and source records attributed to Ewing, R..

5 recordsLinked to original sources

bioRxiv2026

GAE-Δ: A Graph-Learning Framework for Gene Network Rewiring and Clinical Outcome Prediction from Multi-Omics Data

Cancer progression and outcomes are driven in part by changes to molecular networks that result from genetic and/or environmental perturbations. These network changes manifest across multiple interconnected network layers and include accumulation of somatic mutations, altered proteinprotein interactions and dysregulated gene-expression. Here we describe a graph autoencoder-based framework (Graph Autoencoder-Delta (GAE-{Delta})), for characterizing phenotype-specific gene role shifts across multi-omics data. Given samples stratified into two contrasting phenotypic groups and a prior gene interaction network, GAE-{Delta} constructs group-specific gene graphs for each omics modality and trains, for each modality, a single graph autoencoder jointly on both group graphs, so that the two group-conditional embeddings share a common latent space. Contrasting these embeddings defines a multi-omics embedding-shift representation for each gene that reflects how its network role reorganizes across phenotypic contexts. These gene-level shifts are subsequently used for unsupervised gene prioritization, multi-omics late fusion and sample-level classification. Applied to five TCGA cancer types with a survival endpoint, GAE-{Delta} achieves competitive or superior predictive performance compared with classical network-based methods and multi-omics matrix-factorisation methods (MOFA+, iNMF), with statistically significant AUC gains over MOFA+ in three of five cohorts and statistical ties on the remaining two. Beyond predictive performance, the consensus shift genes are significantly enriched for known cancer drivers in three of five cohorts (hypergeometric p < 0.01; 11-17x fold-enrichment), whereas matrix-factorisation baselines reach p < 0.05 in zero of five cohorts (best per-cancer p = 0.06), indicating that GAE-{Delta} captures biological signal that linear factor methods miss. In summary, the GAE-{Delta} approach provides for both improved outcome classification as well as for biological and mechanistic discovery through deep network-based integration of disease-associated multi-omics data.

bioinformatics↗

bioRxiv2025

sampleclusteR: A lightweight R package for automated clustering of transcriptomics samples using metadata

BackgroundAs technologies for genome-wide gene-expression analysis continue to develop, the databases storing the resulting data have grown accordingly. The Gene Expression Omnibus (GEO) has grown to over 250,000 data series across more than 25,000 omics platforms. Likewise, ArrayExpress is comprised of over 70,000 transcriptome and methylome datasets. Conducting meta-analyses of data from these databases can be challenging, typically requiring extensive manual grouping of samples to identify experimental groups for comparison. ResultsHere we present sampleclusteR, a lightweight R package which automates the clustering of gene-expression study samples based on their metadata. To demonstrate the utility of the approach to large scale analysis of GEO data series, 275 GEO data series were analysed using the package. sampleclusteR was able to correctly cluster 4694 of the 5081 samples across the 275 data sets in an unsupervised manner. In addition, 250 datasets from ArrayExpress were analysed by the package with 8547 of the 9154 samples being automatically clustered into correct groups. We show how sampleclusteR can be used to automate analysis of gene-expression datasets by conducting a meta-analysis of multiple GEO data series related to the Wnt signaling pathway. sampleclusteR correctly assigned all samples to the correct experimental groups and identified sets of differentially expressed genes for downstream analysis. ConclusionssampleclusteR enables large-scale analysis of data from GEO or ArrayExpress by automating the clustering of both GEO and ArrayExpress metadata tables using text mining of their associated metadata.

bioinformatics↗

bioRxiv2021

Pseudohypoxic HIF pathway activation dysregulates collagen structure-function in human lung fibrosis

Extracellular matrix (ECM) stiffening with downstream activation of mechanosensitive pathways is strongly implicated in fibrosis. We previously reported that altered collagen nanoarchitecture is a key determinant of pathogenetic ECM structure-function in human fibrosis (Jones et al., 2018). Here, through human tissue, bioinformatic and ex vivo studies we show that hypoxia-inducible factor (HIF) pathway activation is a critical pathway for this process regardless of oxygen status (pseudohypoxia). Whilst TGF{beta} increased rate of fibrillar collagen synthesis, HIF pathway activation was required to dysregulate post-translational modification of fibrillar collagen, promoting bone-type cross-linking, altering collagen nanostructure, and increasing tissue stiffness. In vitro, knock down of Factor Inhibiting HIF (FIH) or oxidative stress caused pseudohypoxic HIF activation in normal fibroblasts. In contrast, endogenous FIH activity was reduced in fibroblasts from patients with lung fibrosis in association with significantly increased normoxic HIF pathway activation. In human lung fibrosis tissue, HIF mediated signalling was increased at sites of active fibrogenesis whilst subpopulations of IPF lung mesenchymal cells had increases in both HIF and oxidative stress scores. Our data demonstrate that oxidative stress can drive pseudohypoxic HIF pathway activation which is a critical regulator of pathogenetic collagen structure-function in fibrosis.

cell biology↗

bioRxiv2018

iOmicsPASS: a novel method for integration of multi-omics data over biological networks and discovery of predictive subnetworks

We developed iOmicsPASS, an intuitive method for network-based multi-omics data integration and detection of biological subnetworks for phenotype prediction. The method converts abundance measurements into co-expression scores of biological networks and uses a powerful phenotype prediction method adapted for network-wise analysis. Simulation studies show that the proposed data integration approach considerably improves the quality of predictions. We illustrate iOmicsPASS through the integration of quantitative multi-omics data using transcription factor regulatory network and protein-protein interaction network for cancer subtype prediction. Our analysis of breast cancer data identifies network signatures surrounding established markers of molecular subtypes. The analysis of colorectal cancer data highlights a protein interactome surrounding key proto-oncogenes as predictive features of subtypes, rendering them more biologically interpretable than the approaches integrating data without a priori relational information. However, the results indicate that current molecular subtyping is overly dependent on transcriptomic data and crude integrative analysis fails to account for molecular heterogeneity in other -omics data. The analysis also suggest that tumor subtypes are not mutually exclusive and future subtyping should therefore consider multiplicity in assignments.\n\nAvailability: https://github.com/cssblab/iOmicsPASS

systems biology↗

bioRxiv2017

The PTPRT pseudo-phosphatase domain is a denitrase

Protein tyrosine nitration occurs under both physiological and pathological conditions1. However, enzymes that remove this protein modification have not yet been identified. Here we report that the pseudo-phosphatase domain of protein tyrosine receptor T (PTPRT) is a denitrase that removes nitro-groups from tyrosine residues in paxillin. PTPRT normally functions as a tumor suppressor and is frequently mutated in a variety of human cancers including colorectal cancer2,3. We demonstrate that some of the tumor-derived mutations located in the pseudophosphatase domain impair the denitrase activity. Moreover, PTPRT mutant mice that inactivate the denitrase activity are susceptible to carcinogen-induced colon tumor formation. This study uncovers a novel enzymatic activity that is involved in tumor suppression.

cancer biology↗