bioRxiv

Model sorts disease-linked changes in cell data by aging pattern

Using cross-sectional single-cell cohorts, scCody distinguished disease-associated cellular changes that resembled healthy aging from changes that departed from it.

Paper by Jiramet Kinchagawat, Jinmiao Chen, Sai Zhang, Xiaotao Shen · Nanyang Technological University
Branching dark blue-green and gray neurons face each other above immune cells, clustered densely on one side and sparsely on the other.

Our read

AI & data
Studied:
Preprint: computational study of cell populations from single-cell cohorts
What changed:
Selected Alzheimer's nerve cells showed aging-aligned changes; COVID-19 immune cells showed weaker alignment and more departures from healthy aging
The catch:
Actual cellular changes as individual donors aged were not measured

Editor of record: Boris Djordjevic · AgingResearch.News

Follow @agingbrief

Cell populations from cross-sectional single-cell cohorts were analyzed with scCody, a computational model that learned how their makeup changed with age. Unlike biological clocks that summarize molecular changes as one age estimate, it compared disease-associated changes with a learned reference for healthy aging.

The model reconstructed cell populations at ages it had not seen during training. In Alzheimer's disease, selected nerve-cell populations showed substantial changes aligned with healthy aging, alongside shared departures associated with established disease-related molecular patterns. COVID-19 showed weaker alignment with aging and more pronounced immune-cell changes that departed from it. These comparisons separated changes resembling healthy aging from those following a different pattern.

Why it matters

Molecular measures of biological age can obscure how disease relates to healthy aging. The authors propose extending age assessment beyond a single number to address that distinction.

Caveats

The model inferred age patterns by comparing different donors, not by tracking the same donors as they aged. The findings come from a preprint that has not been peer-reviewed.

Comments

Be the first to comment.