Abstract Motivation Accurate classification of patients into molecular subgroups is critical for the development effective therapeutics and deciphering what drives these to cancer. The availability multiomics data catalogs large cohorts cancer provides multiple views biology tumors with unprecedented resolution. Results We develop Pathway-based MultiOmic Graph Kernel clustering (PAMOGK) that integrates patient existing biological knowledge on pathways. a novel graph kernel evaluates...

Abstract Accurate classification of patients into molecular subgroups is critical for the development effective therapeutics and deciphering what drives these to cancer. The availability multi-omics data cat-alogs large cohorts cancer provides multiple views biology tumors with unprecedented resolution. We develop PAMOGK ( Pa thway based M ulti O mic G raph K ernel clustering) that not only integrates patient existing biological knowledge on pathways. a novel graph kernel evaluates...

Anti-cancer therapies based on synthetic lethality (SL) exploit tumour vulnerabilities for treatment with reduced side effects, by targeting a gene that is jointly essential another whose function lost. Computational prediction key to expedite SL screening, yet existing methods are vulnerable prevalent selection bias in data and reliant cancer or tissue type-specific omics, which can be scarce. Notably, sequence similarity remains underexplored as proxy related joint essentiality.

Abstract Motivation Synthetic lethality (SL) between two genes occurs when simultaneous loss of function leads to cell death. This holds great promise for developing anti-cancer therapeutics that target synthetic lethal pairs endogenously disrupted genes. Identifying novel SL relationships through exhaustive experimental screens is challenging, due the vast number candidate pairs. Computational prediction therefore sought identify promising gene further experimentation. However, current...

Background: Soil naturally harbours antibiotic resistant bacteria and is considered to be a reservoir for resistance. The overuse of antibiotics across human, animal, environmental sectors has intensified this issue leading an increased acquisition genes by in soil. Various biogeographical factors, such as soil pH, temperature, pollutants, play role the spread emergence resistance In study, we utilised publicly available metagenomic datasets from four different types (rhizosphere, urban,...

Fairness in machine learning seeks to mitigate model bias against individuals based on sensitive features such as sex or age, often caused by an uneven representation of the population training data due selection bias. Notably, unascribed is challenging identify and typically goes undiagnosed, despite its prominence complex high-dimensional from fields like computer vision molecular biomedicine. Strategies unidentified evaluate mitigation methods are crucially needed, yet remain...

<title>Abstract</title> <bold>Background:</bold> Soil naturally harbours antibiotic resistant bacteria and is considered to be a reservoir for resistance. The overuse of antibiotics across human, animal, environmental sectors has intensified this issue leading an increased acquisition genes by in soil. Various biogeographical factors, such as soil pH, temperature, pollutants, play role the spread emergence resistance In study, we utilised publicly available metagenomic datasets from four...

Abstract Motivation Many tumours show deficiencies in DNA damage response (DDR), which influence tumorigenesis and progression, but also expose vulnerabilities with therapeutic potential. Assessing patients might benefit from DDR-targeting therapy requires knowledge of tumour DDR deficiency status, mutational signatures reportedly better predictors than loss function mutations select genes. However, are identified independently using unsupervised learning, therefore not optimised to...

Selection bias poses a critical challenge for fairness in machine learning, as models trained on data that is less representative of the population might exhibit undesirable behavior underrepresented profiles. Semi-supervised learning strategies like self-training can mitigate selection by incorporating unlabeled into model training to gain further insight distribution population. However, conventional seeks include high-confidence samples, which may reinforce existing and compromise...

A gene is considered essential if its function indispensable for the viability or reproductive success of a cell an organism. Distinguishing genes from non-essential ones fundamental question in genetics, and it key to understanding minimal set functional requirements Knowledge also crucial drug discovery. Several reports literature show that location protein-protein interaction network correlated with target gene’s essentiality. Here, we ask whether node embeddings (PPI) can help predict...

Abstract Anti-cancer therapies based on synthetic lethality (SL) exploit tumor vulnerabilities for treatment with reduced side effects. Since simultaneous loss-of-function of SL genes causes cell death, tumors known gene disruptions can be treated by targeting partners. Computational selection promising candidates amongst all combinations is key to expedite experimental screening. However, current prediction models: (i) only use tissue type-specific molecular data, which scarce/noisy,...