Projects

During cancer development, cancer cells grow into multiple genetically related sub-lineages that evolve in parallel and display heterogeneity at multiple levels. This project aims to trace, identify and isolate drug resistant sub-lineages of cancer cell line and PDX models using a cutting-edge DNA barcoding platform developed in the group. By utilizing CRISPR activation, sub-lineages of interest can be isolated in a highly-efficiently manner, which is not feasible with conventional DNA barcoding systems. Isolated cancer sub-lineages can be then biologically characterized using multi-omics profiling to identify key features in drug resistant sub-lineages. The project will improve our understanding of drug-resistant cancer cells and facilitate the development of therapeutic interventions to tackle treatment resistance.

Even if targeted therapies have revolutionized the treatment of many cancers, most patients develop resistance, have severe side-effects or relapse during treatment. Therefore, there is a clinical need to tailor optimal therapy for each patient to prevent ineffective treatment and toxic effects.  This project will identify multi-marker panels and implement AI-based clinical decision support systems to guide treatment decisions for individual patients. We expect this will lead to increased treatment efficacy, individualization of therapy and reduced drug use and side effects. In addition, the multi-omics guided predictions will improve future trial designs and biomarker-based therapy selection.

 https://cll-clue.eu/

This multidisciplinary project develops, tests and implements novel practices of how to use artificial intelligence (AI) and machine learning (ML) models in translational and clinical studies. The modeling goal is to develop novel supervised learning approaches to select multi-omic features predictive of clinical outcomes for individual patients using efficient AI / ML models that maximize the accuracy of the predicted outcome, using minimal panel of predictive features. The medical question is how to improve treatment outcomes, and how to do this cost-effectively to minimize the burden on public health expenditure.

Zhao et al. iScience 2022; https://doi.org/10.1016/j.isci.2022.104767

Personalized medicine faces critical challenges in terms of the diagnosis, prognosis and treatment response prediction of patients with pancreatic cancer. Projected to be the second most lethal cancer by the end of 2030, and with the lowest survival rate across all cancers in Europe, pancreatic cancer calls for a joint action across scientific disciplines and institutions, both in industry and academia. PANCAIM is a pan-European project aiming to combine AI expertise with clinicians’ and biologists’ knowledge to develop an AI-guided platform that will allow the integration of multiple types of cancer data (from genomics to medical images) and their integrated analysis using multi-modal predictive models. The PANCAIM platform is expected to become not only a centralized knowledge hub for pancreatic cancer biology but also a repository of accountable, AI-assisted clinical applications, that will help clinicians to select the more promising treatment options for pancreatic cancer patients.

https://pancaim.eu/