Unlocking the Potential of Omics: A Translational Approach to Precision Medicine with Deep Learning
Description: Our project´s objective is to revolutionize precision medicine by integrating deep learning with extensive omics datasets and protein interaction networks, thereby enhancing the accuracy and usability of disease diagnostics and therapeutic strategies. We begin by exploring vast single omics databases, integrating these with the protein interaction network through special types of deep autoencoders. Our aim is to establish a new cellular representation, driven by data and extensive knowledge, that defines a unified space for various omics types. This representation will then be refined using paired omic collections to construct translational neural networks, laying the groundwork for the identification of biomarkers across omics. Furthermore, these networks will aid in developing and enhancing interpretable classifiers for more than seven complex risk statuses and diseases. For instance, our work on multi-omic biological age calculators could significantly benefit forensic sciences, which have traditionally relied on a narrow range of markers, thus ignoring the potential of utilizing millions of features. In summary, this project, supported by a multidisciplinary team and leveraging extensive datasets, is poised at the cutting edge of incorporating advanced computational methods into healthcare, striving to make personalized medicine a practical reality for everyone.
Funder: Vetenskapsrådet
Amount: SEK 4,000,000
Funding type: research
Funder scheme: Project grant
Start date: 2025-01-01
End date: 2028-12-31
Investigator: Mika Gustafsson
Institution: Linköping University
Funded outputs count: 0
Funder award ID: 2024-05815
OpenAlex ID: G4081240353