Biology subjects

Bessant, C.

Publications and source records attributed to Bessant, C..

5 recordsLinked to original sources

bioRxiv2025

MORGaN: self-supervised multi-relational graph learning for druggable gene discovery

Identifying therapeutically tractable targets remains difficult, partly because disease biology is distributed across multiple molecular layers and relation types, while labeled data are scarce. We present MORGaN, a self-supervised framework for node classification on multi-omic, multi-relation gene networks that learns structure-aware embeddings and outputs calibrated scores to prioritize therapeutic targets. On a pan-cancer graph integrating TCGA multi-omics and diverse biological relation types, MORGaN outperforms state-of-the-art biological node classification models across metrics (AUPR: 0.815 [->] 0.888; +9%). Ablation studies highlight that both relation diversity and the in-layer fusion architecture are necessary for these gains. Prioritized targets are biologically coherent: high-confidence hits are enriched for pharmaceutically tractable families and ligand-receptor signaling cascades. Post hoc explainability analyses recover compact, pathway-consistent motifs around both known and putative novel targets, and concordance with external resources further supports plausibility. MORGaN thus delivers label-efficient, interpretable node classification for target discovery and can be readily adapted to other diseases, other species, and other node classification tasks. Code and documentation are available at this link.

bioinformatics↗

bioRxiv2024

Understanding the Sources of Performance in Deep Learning Drug Response Prediction Models

Anti-cancer drug response prediction (DRP) using cancer cell lines plays a vital role in stratified medicine and drug discovery. Recently there has been a surge of new deep learning (DL) models for DRP that improve on the performance of their predecessors. However, different models use different input data types and neural network architectures making it hard to find the source of these improvements. Here we consider multiple published DRP models that report state-of-the-art performance in predicting continuous drug response values. These models take the chemical structures of drugs and omics profiles of cell lines as input. By experimenting with these models and comparing with our own simple benchmarks we show that no performance comes from drug features, instead, performance is due to the transcriptomics cell line profiles. Furthermore, we show that, depending on the testing type, much of the current reported performance is a property of the training target values. To address these limitations we create novel models (BinaryET and BinaryCB) that predict binary drug response values, guided by the hypothesis that this reduces the noise in the drug efficacy data. Thus, better aligning them with biochemistry that can be learnt from the input data. BinaryCB leverages a chemical foundation model, while BinaryET is trained from scratch using a transformer-type model. We show that these models learn useful chemical drug features, which is the first time this has been demonstrated for multiple DRP testing types to our knowledge. We further show binarising the drug response values is what causes the models to learn useful chemical drug features. We also show that BinaryET improves performance over BinaryCB, and over the published models that report state-of-the-art performance.

bioinformatics↗

bioRxiv2022

Community detection in empirical kinase networks identifies new members of signaling pathways

Phosphoproteomics allows one to measure the activity of kinases that drive the fluxes of signal transduction pathways involved in biological processes such as immune function, senescence and growth. However, deriving knowledge of signaling network circuitry from these data is challenging due to a scarcity of phosphorylation sites that define kinase-kinase relationships. To address this issue, we previously identified around 6,000 phosphorylation sites markers of kinase-kinase relationships (that may be conceptualised as network edges), from which empirical cell-model-specific weighted kinase networks may be reconstructed. Here, we assess whether the application of community detection algorithms to such networks can identify new components linked to canonical signaling pathways.Phosphoproteomics data from acute myeloid leukaemia (AML) cells treated separately with PI3K, ATK, MEK and ERK inhibitors were used to reconstruct individual kinase networks. In each network, we applied the community detection method modularity maximization and selected the community containing the main target of the inhibitor the cells were treated with. These analyses returned communities that contained known canonical signaling components. Interestingly, in addition to canonical PI3K/AKT/MTOR members, the community assignments returned TTK (also known as MPS1) as a likely component of PI3K/AKT/mTOR signaling. We confirmed this observation with wet-lab laboratory experiments showing that TTK phosphorylation was decreased in AML cells treated with AKT and MTOR inhibitors. This study illustrates the application of community detection algorithms to the analysis of empirical kinase networks to uncover new members linked to canonical signaling pathways. Author summaryKinases are key enzymes that regulate the transduction of extracellular signals from cell surface receptors to changes in gene expression via a set of kinase-kinase interactions and signalling cascades. Inhibiting hyperactive kinases is a viable therapeutic strategy to treat different cancer types. Unfortunately, kinase signalling networks are robust to external perturbations, thus allowing tumour cells to orchestrate mechanisms that compensate for inhibition of specific kinases. Therefore, there is a need to better understand kinase network structure and to identify new therapeutic targets. Here, we reconstructed kinase networks from phosphoproteomics data, and compared the activity of its kinase interactions in acute myeloid leukaemia (AML) cells. We then tested community detection algorithms to identify kinase components associated to PI3K/AKT/MTOR signalling, a paradigmatic oncogenic signalling cascade. We found that TTK was usually grouped with networks derived for PI3K, AKT and MTOR kinases. Wet-lab experiments confirmed that TTK is likely to act downstream of AKT and MTOR. We thus show that our methods can be used to identify potential new members of canonical kinase signalling cascades.

systems biology↗

bioRxiv2022

HNRNPA2B1 controls an unfolded protein response-related prognostic gene signature in prostate cancer

HNRNPA2B1 is associated with prostate cancer (PC) disease aggressiveness and underlies pro-tumourigenic cellular stress responses. By analysing >500 PC transcriptomes, we reveal that HNRNPA2B1 over-expression is associated with poor patient prognosis and stress response pathways. These include the "protein processing in the endoplasmic reticulum" (ER) pathway, which incorporates the unfolded protein response (UPR). By RNA-sequencing of HNRNPA2B1-depleted cells PC cells, we identified HNRNPA2B1-mediated down-regulation of UPR genes including the master ER-stress sensor IRE1, which induces ER proteostasis. Consistent with IRE1 down-regulation in HNRNPA2B1-depleted cells, we observed reduced splicing of the IRE1-target and key UPR effector XBP1s. Furthermore, HNRNPA2B1 depletion up-regulates expression of the IRE1-dependent decay (RIDD) target gene BLOC1S1, which is degraded by activated IRE1. We identify a HNRNPA2B1-IRE1-XBP1-controlled four gene prognostic biomarker signature (HIX) which classifies a subgroup of primary PC patients at high risk of disease relapse. Pharmacological targeting of IRE1 attenuated HNRNAPA2-driven PC cell growth. Taken together, our data reveal a putative novel mechanism of UPR activation in PC by HNRNPA2B1, which may promote an IRE1-dependent yet potentially-targetable recurrent disease phenotype.

cancer biology↗

bioRxiv2017

GeoDiver: Differential Gene Expression Analysis & Gene-Set Analysis For GEO Datasets

SummaryGeoDiver is an online web application for performing Differential Gene Expression Analysis (DGEA) and Generally Applicable Gene-set Enrichment Analysis (GAGE) on gene expression datasets from the publicly available Gene Expression Omnibus (GEO). The output produced includes numerous high quality interactive graphics, allowing users to easily explore and examine complex datasets instantly. Furthermore, the results produced can be reviewed at a later date and shared with collaborators.\n\nAvailabilityGeoDiver is freely available online at http://www.geodiver.co.uk. The source code is available on Github: https://github.com/GeoDiver/GeoDiver and a docker image is available for easy installation.

bioinformatics↗