WP7 – Bioinformatics

The AwE bioinformatics will integrate consortium data and public information to infer healthspan-related pathways and biomarkers, and to suggest compounds. Fusion of chemical, pharmaceutical, clinical and multi-omics data, including genetic data, RNAi, gene and protein expression, and metabolomics will be performed at multiple levels. We take gene/protein interaction as the universal reference network, and attach genetic/RNAi and metabolic information as appropriate, following the network-based link-score highlighting idea we pioneered (ExprEssence). Scoring of interactions allows feature reduction for machine learning of healthspan determinants and compound synergy, as well as for better mechanistic understanding. At the level of similarities and raw data, we apply kernel methods for large-scale fusion in predictive tasks and Bayesian approaches over probabilistic graphical models (References) to infer robust biomarkers and causal relations.