Project Overview

The increased availability of high-throughput technologies has enabled the generation and collection of different modalities of high-dimensional molecular data that attempt to map many different but complementary biological layers, including genomics, epigenomics, transcriptomics, proteomics, lipidomics, and metabolomics. To study complex biological processes holistically, it is imperative to adopt an integrative approach that combines different types of multi-omics data to provide a comprehensive view of the mechanisms of disease or biological processes. Understanding molecular behavior, pathway interactions, and relationships between and within the heterogeneous data types could improve disease diagnosis, prognosis, and treatment.

However, integrating multi-omics datasets poses several challenges, including their high dimensionality, limited number of patients, heterogeneity, and modeling interactions between the different types of omics data. As a result, computational mechanisms for data analysis must be continuously adapted to analyze high-dimensional and heterogeneous omics data more accurately and efficiently.

The project aims to develop effective and novel computational methods for multi-omics data integration, which combine statistical methods, optimization approaches, and matrix factorization, leveraging cutting-edge research results to which we have actively contributed. Developed methodologies will be released as open-source packages (considering the critical concepts of scalability and medical interpretability) and embedded into larger pipelines to analyze a few case studies from cell development and cancers.
Overall, this project is expected to impact the strategic topic of HUMAN WELLBEING by providing algorithms that can unlock the potential of these new technologies toward precision medicine.

Although the proposed project is motivated in a multi-omics data context, it will enable scientific advantages in the more general area of mathematical tools for integrating large and heterogeneous data collected from different sources and instruments. This will also contribute to advancing the mathematical framework behind the data integration theory proposing more rigorous and efficient computational approaches.

CAIMOD Project: Computational Approaches for Integration of Multi-Omics Data

CUP: H53D23008870001

Finanziato dall’Unione europea – Next Generation EU