Capturing data…
Capturing data…
TUE, 27 OCT · 97 ITEMS
"Blood-trained protein EpiScores show limited cross-tissue concordance in GTEx" — bioRxiv (Genomics) · Science & Medicine
This is a preprint study testing whether 'EpiScores' — DNA methylation-based scores trained on blood to predict blood protein levels — actually hold up when applied to other tissues. Using the dataset, the authors compared blood-trained protein EpiScores against measured protein levels in matched non-blood tissues and found limited concordance, meaning the scores don't transfer well across tissues. The implication is that blood-derived epigenetic protein predictions may not be reliable proxies for protein levels in other tissues, which matters for studies using such scores as biomarkers.
The claim is based directly on the preprint's own analysis of GTEx data, so it is solid as a report of the authors' findings. However, as a preprint, it has not yet undergone peer review, and the specific numbers and methods are not detailed in the headline; the conclusion of 'limited concordance' is the authors' interpretation of their results.
The preprint was published on bioRxiv on 2026-10-26, so this is genuinely new research as of that date.