Search

Neoantigen immunogenicity prediction is a highly challenging problem in the development of personalised medicines. Low reactivity rates in called neoantigens result in a difficult prediction scenario with limited training datasets. Here we describe ImmugenX, a modular protein language modelling approach to immunogenicity prediction for CD8+ reactive epitopes. ImmugenX comprises of a pMHC encoding module trained on three pMHC prediction tasks, an optional TCR encoding module and a set of context specific immunogenicity prediction head modules. Compared with state-of-the-art models for each task, ImmugenX's encoding module performs comparably or better on pMHC binding affinity, eluted ligand prediction and stability tasks. ImmugenX outperforms all compared models on pMHC immunogenicity prediction (Area under the receiver operating characteristic curve = 0.619, average precision: 0.514), with a 7% increase in average precision compared to the next best model. ImmugenX shows further improved performance on immunogenicity prediction with the integration of TCR context information. ImmugenX performance is further analysed for interpretability, which locates areas of weakness found across existing immunogenicity models and highlight possible biases in public datasets. Author summary: Accurate prediction of neoantigen immunogenicity has the potential to greatly improve the effectiveness of targeted therapies for cancer. While there are a number of associated tasks, such as peptide-HLA elution prediction, which can be now predicted with high accuracy by a number of published models, direct prediction of which epitopes will produce an immune response has proven more difficult. In this paper we demonstrate a modular protein language model approach which is trained iteratively to include data from related sub-tasks and can be extended to include information such as candidate TCRs of interest when available. There is a relatively small amount of immunogenicity data available, with even less data available with paired TCRs. This makes directly training a model to predict immunogenicity challenging. Our approach has the advantage of utilising data from sub-tasks and masked language modelling to allow for training a highly performant model with a small dataset. Using a cancer-specific benchmarking dataset we show this approach improves on existing state-of-the-art models and can be improved further with the addition of TCR context. This provides a framework that can serve as the basis for utilising additional information sources and datasets as they become available. [ABSTRACT FROM AUTHOR]

Biological pathways between alcohol consumption and alcohol liver disease (ALD) are not fully understood. We selected genes with known effect on (1) alcohol consumption, (2) liver function, and (3) gene expression. Expression of the orthologs of these genes in Caenorhabditis elegans and Drosophila melanogaster was suppressed using mutations and/or RNA interference (RNAi). In humans, association analysis, pathway analysis, and Mendelian randomization analysis were performed to identify metabolic changes due to alcohol consumption. In C. elegans, we found a reduction in locomotion rate after exposure to ethanol for RNAi knockdown of ACTR1B and MAPT. In Drosophila, we observed (1) a change in sedative effect of ethanol for RNAi knockdown of WDPCP, TENM2, GPN1, ARPC1B, and SCN8A, (2) a reduction in ethanol consumption for RNAi knockdown of TENM2, (3) a reduction in triradylglycerols (TAG) levels for RNAi knockdown of WDPCP, TENM2, and GPN1. In human, we observed (1) a link between alcohol consumption and several metabolites including TAG, (2) an enrichment of the candidate (alcohol-associated) metabolites within the linoleic acid (LNA) and alpha-linolenic acid (ALA) metabolism pathways, (3) a causal link between gene expression of WDPCP to liver fibrosis and liver cirrhosis. Our results imply that WDPCP might be involved in ALD. [ABSTRACT FROM AUTHOR]

Alcohol consumption is associated with the development of cardiovascular diseases, cancer, and liver disease. The biological mechanisms are still largely unclear. Here, we aimed to use an agnostic approach to identify phenotypes mediating the effect of alcohol on various diseases. Methods: We performed an agnostic association analysis between alcohol consumption (red and white wine, beer/cider, fortified wine, and spirits) with over 7800 phenotypes from the UK biobank comprising 223,728 participants. We performed Mendelian randomisation analysis to infer causality. We additionally performed a Phenome-wide association analysis and a mediation analysis between alcohol consumption as exposure, phenotypes in a causal relationship with alcohol consumption as mediators, and various diseases as the outcome. Results: Of 45 phenotypes in association with alcohol consumption, 20 were in a causal relationship with alcohol consumption. Gamma glutamyltransferase (GGT; β = 9.44; 95% CI = 5.94, 12.93; Pfdr = 9.04 × 10−7), mean sphered cell volume (β = 0.189; 95% CI = 0.11, 0.27; Pfdr = 1.00 × 10−4), mean corpuscular volume (β = 0.271; 95% CI = 0.19, 0.35; Pfdr = 7.09 × 10−10) and mean corpuscular haemoglobin (β = 0.278; 95% CI = 0.19, 0.36; Pfdr = 1.60 × 10−6) demonstrated the strongest causal relationships. We also identified GGT and physical inactivity as mediators in the pathway between alcohol consumption, liver cirrhosis and alcohol dependence. Conclusion: Our study provides evidence of causality between alcohol consumption and 20 phenotypes and a mediation effect for physical activity on health consequences of alcohol consumption. [ABSTRACT FROM AUTHOR]