Data from: Joint variable selection for omic biomarkers in time-to-event data
Data files
May 13, 2026 version files 2.69 MB
Abstract
The incidence of the vast majority of neurodegenerative, cancer, and metabolic diseases generally increases exponentially with age. In large-scale biobanks, linking time-to-diagnosis information in electronic health records to multiple genomic (“multiomics”) measures has the potential to reveal the genes and biological pathways involved in the disease onset and progression. To date, association testing has commonly been conducted by testing one variable at a time, which ignores correlation structure and increases the risk of false discoveries. To address these issues, we introduce a novel fully parametric Bayesian computational method, vampW, based on the Vector Approximate Message Passing framework applied to a Weibull model. vampW jointly models correlated features, provides joint association testing via joint Posterior Inclusion Probabilities (PIPs), and incorporates prior knowledge. Here, we report PIPs obtained from the analysis of 53,018 UK Biobank participants across 24 disease outcomes. Using a 95% PIP threshold, vampW identifies 219 protein-disease associations. After correcting protein levels for exponential age effects in addition to linear age and sex correction, vampW identifies 1,308 associations. The findings replicate in independent cohorts using different measurement technologies, within data from Iceland and a novel Generation Scotland proteomics dataset.
Dataset DOI: 10.5061/dryad.9ghx3fg01
This dataset contains summary statistics from the study “Joint Variable Selection for Omic Biomarkers in Time-to-Event Data.” In the study, a new survival analysis method called vampW was developed and applied to UK Biobank proteomics data.
vampW is a Bayesian joint regression method that estimates Posterior Inclusion Probabilities (PIPs) for biomarkers. A PIP quantifies the probability that a biomarker is associated with a given disease outcome after accounting for all other biomarkers in the model.
Description of the Data and File Structure
The dataset consists of two tabular files. In each file:
- rows correspond to protein biomarkers,
- columns correspond to 24 disease outcomes from the UK Biobank,
- entries contain the estimated PIPs.
Before analysis, protein levels were adjusted for covariates (age and sex) using two different correction models, resulting in two output files:
- vampW_posterior_inclusion_probabilities.csv
Protein levels adjusted for age and sex using a linear correction model. - vampW_posterior_inclusion_probabilities_with_exponential_age_correction.csv
Protein levels adjusted for age and sex using a linear correction model with an additional exponential age term.
Code/software
The reported data were inferred using the vampW software. The description of the model, source code, requirements, and information on how to set up the code are available in the repository: https://github.com/Information-and-learning-for-genomics/Time2EVAMP