OMICSPRED is a resource for predicting multi-omics data (proteomics, metabolomics, transcriptomics etc.) directly from genotypes. To do this, we have used a single cohort (INTERVAL) with extensive multi-omics data to train genetic scores using machine learning. Here, you can explore and download the genetic scores for a wide range of biomolecular traits in human blood as well as the summary statistics of their associations with key traits and diseases in the UK Biobank.
Genetic scores were trained on the INTERVAL cohort using Bayesian Ridge regression. Validation was performed on independent individuals from other cohorts or on withheld subsets of INTERVAL (more info below). Detailed methods and validation steps can be found here.
Platforms with Genetic Scores
Application of Genetic Scores
OMICSPRED is under active development, and we will continue to add genetic scores. If you use OMICSPRED in your research, we ask that you cite our submitted ASHG 2021 abstract for now (below). A full manuscript is in preparation and is anticipated to be preprinted in Q3/Q4 2021.
Manuscript: Xu Y. et al. An atlas of genetic scores to predict multi-omic biomolecular traits in blood. (in preparation)
ASHG Abstract: Yu Xu, Scott Ritchie, Maik Pietzner, Samuel Lambert, Sebastian May-Wilson, Artika Nath, Praveen Surendran, Åsa Johansson, Elodie Persyn, Loïc Lannelongue, Bram Prins, Nicola Pirastu, Dirk Paul, Christopher Yau, James F. Wilson, Claudia Langenberg, Anders Mälarstig, John Danesh, Adam Butterworth, Michael Inouye. An atlas of genetic scores to predict multi-omic biomolecular traits in blood. American Society of Human Genetics (2021)
Questions and Feedback
We would love to hear from you! To provide feedback or ask a question, you can contact the OMICSPRED team here.