Meta Analyses of Heterogeneous Omics Data

Informazioni relative al progetto

MAHOD

ID dell’accordo di sovvenzione: 101106986

DOI

10.3030/101106986

Progetto concluso il 31 Maggio 2026

Data della firma CE 10 Luglio 2023

Data di avvio 18 Marzo 2024

Data di completamento 17 Giugno 2026

Finanziato da

• Marie Skłodowska-Curie Actions (MSCA)

Costo totale

Nessun dato

Contributo UE

€ 181 152,96

Coordinato da

UNIVERSIDAD DE GRANADA
Spain

Periodic Reporting for period 1 - MAHOD (Meta Analyses of Heterogeneous Omics Data)

Periodo di rendicontazione: 2024-03-18 al 2026-06-17

Sintesi del contesto e degli obiettivi generali del progetto

Omics experiments are commonly disseminated as tabular datasets containing relatively few biological samples but thousands of measured variables. This combination may make statistical estimates unstable requiring many parallel hypothesis tests that increase the risk of false discoveries, or apparent differences between experimental groups that arise through random variation rather than a reproducible biological effect. Such findings can direct subsequent research towards hypotheses that are not genuinely promising.

Traditional meta-analysis combines evidence from multiple studies by comparing effect estimates for the same or sufficiently similar measurements, such as the abundance of a shared clinical marker. However, many potentially related omics studies contain neither the same samples nor the same measured variables and therefore cannot be combined using conventional approaches. Meta Analyses of Heterogeneous Omics Data (MAHOD) aims to develop a mathematical and statistical framework for evaluating such datasets by exploiting their common experimental design. For example, comparable treatment and control groups, even when their observations and measurement modalities do not overlap.

Lavoro eseguito dall’inizio del progetto fino alla fine del periodo coperto dalla relazione e principali risultati finora ottenuti

This project investigated methods for analysing heterogeneous datasets in which features were not directly comparable between studies, as well as heteromodal datasets obtained using entirely different measurement technologies. It also furthered a simulation-based method for estimating the number of study participants required to achieve a specified level of statistical power under relatively mild assumptions.

Progressi oltre lo stato dell’arte e potenziale impatto previsto (incluso l’impatto socioeconomico e le implicazioni sociali più ampie del progetto fino ad ora)

MAHOD demonstrated that, for certain types of chemical instrumentation, transforming signals into the frequency domain can make statistical inference more robust to drift occurring in the time domain. It also established conditions under which datasets obtained using entirely different measurement modalities can be analysed jointly through a transformation with a unique solution. Methods were evaluated for assessing agreement between these datasets while accounting for possible ambiguities in their low-rank representations. More broadly, the project outlined strategies for analysing raw instrumental signals through linear transformations in the frequency domain when an equivalent analysis is not feasible in the time domain.