About this webserver
This webserver provides open access to summary statistics from our work on molecular signatures across 27 incident non-communicable diseases (NCDs) and all-cause mortality for academic purposes.
We used baseline untargeted plasma metabolomics profiling covering 1,014 metabolites using the Metabolon HD4 platform to characterise disease-shared (two-thirds) and disease-specific (one-third) pathways among up to 11,966 participants (219,415 person-years) of the EPIC-Norfolk cohort. We integrated baseline data on over 50 clinical risk factors and traits to put metabolite-disease associations into context.
- Cox-models for incident diseases
- Linear regression models for clinical risk factors
Summary statistics are presented in a matrix format with metabolites as rows and clinical data in columns. Users can filter/subset the matrix based on statistical criteria, such as association strength or effect estimates, or restricted to an exposure or outcome of interest. The format allows to easily investigate specificity of metabolite associations and to enable cross-linkage of exposure/outcome data. Publication-ready plots for candidate metabolites can be saved and summary statistics can be downloaded for the use of meta-analysis.
To facilitate comparability among all estimates we log-transformed, winsorised (mean ± 5 standard deviations), and standardized (mean of zero and standard deviation of one) plasma metabolite levels.
A more detailed explanation of methods and the study could be found in the original publication and associated Supplemental Material.
Cox-proportional hazard models for incident diseases
Exposure: plasma metabolite levels. Outcome: risk of 27 diseases (ICD-10 codes) and all-cause mortality.
Linear regression models for clinical risk factors
Exposure: >50 baseline characteristics. Outcome: plasma metabolite levels.
Any access to summary statistics and other information provided by this webserver is strictly for academic use and not for commercial interest (see Licence terms).
Please contact Dr. Claudia Langenberg for any further information.