Work Programme

  1. Benchmarking environment
    To improve the explainability of the data integration workflow, create benchmarking datasets in two ways: i) experimental (“real life”), and ii) synthetic data. In WP3, the datasets are used to create efficient multi-omic specific analysis pipelines. Hence, synthetic data generated in this work package serve as a gold standard. 
  2. Muti-model framework development for reproducible and explainable multi-omics data integration
    Design and implement a framework for reproducible and explainable multi-omics data integration. Using this framework, investigate how preprocessing, sparsity, and various data transformation techniques influences particular combinations of data-driven and/or knowledge-based approaches, among others. Investigate how the principle of divide and conquer (e.g. when feature extraction precedes integration), influences steps such as classification or clustering. Test robustness of methods to unbalanced data, especially in cases where the number of conditions is higher than number of biological replicates.  
  3. Biological evaluation of the array–of–models within developed framework
    Biological evaluation of the outputs of the arrays of models (developed in WP2) is crucial for ensuring that the computational outcomes align with known biological principles. Use existing biological data (from T1.1) and generate hypotheses for wet-lab validation in order to identify most suitable model combination for experimental scenarios.  
  4. Project management, data management, communication, dissemination
    Data generated within the project will are organized according to FAIR principles for data management. All experiments, both wet lab and dry lab, are recorded in real-time in our data management system pISA-tree.