Back

DE-SWAN analysis of biomarker trajectories over the life course is highly sensitive to age distribution and model parameterization

Che, A.; Shah, P.; Tanner, K. T.; Cohen, A. A.

2026-10-05 bioinformatics
10.64898/2026.09.29.754760 bioRxiv
Show abstract

DE-SWAN is a commonly used method to study how large ensembles of omics biomarkers change across the life course. It has been used to demonstrate apparent repeated accelerations and decelerations of aging rates across the life course, including 2-3 peaks of aging-related changes. This result contradicts what is expected based on the demography of aging and known biological mechanisms. Here, we use simulations to study the robustness of DE-SWAN to the age distribution of the study sample and to choice of window size. We show that under many realistic scenarios, DE-SWAN is highly sensitive to factors unrelated to the underlying rates of change in aging biology. This calls into question the results of previous analyses that have attributed DE-SWAN peaks to acceleration and deceleration of aging. We conclude that the method is unreliable for studying rates of change across the life course under most circumstances.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

1
PLOS Computational Biology
● 1863 papers in training set
Top 1%
19.1%
Similar papers in this journal
2
eLife
● 5828 papers in training set
Top 15%
7.5%
Similar papers in this journal
3
Biogerontology
○ 10 papers in training set
Top 0.1%
7.0%
Similar papers in this journal
4
Scientific Reports
○ 3612 papers in training set
Top 11%
6.5%
Similar papers in this journal
5
BMC Bioinformatics
○ 457 papers in training set
Top 1%
5.7%
Similar papers in this journal
6
Nature Communications
○ 5641 papers in training set
Top 27%
5.7%
50% of probability mass above
7
PLOS ONE
● 5266 papers in training set
Top 30%
4.5%
8
Briefings in Bioinformatics
◐ 354 papers in training set
Top 3%
3.3%
9
GeroScience
○ 109 papers in training set
Top 0.8%
2.9%
10
Proceedings of the National Academy of Sciences
● 2444 papers in training set
Top 21%
2.5%
11
Aging Cell
○ 165 papers in training set
Top 1%
2.2%
12
Cell Systems
○ 201 papers in training set
Top 3%
1.8%
13
Frontiers in Genetics
○ 230 papers in training set
Top 2%
1.7%
14
npj Aging
○ 22 papers in training set
Top 0.2%
1.7%
15
Journal of Theoretical Biology
○ 162 papers in training set
Top 1%
1.5%
16
Aging
○ 75 papers in training set
Top 1%
1.4%
17
Communications Biology
○ 993 papers in training set
Top 20%
1.2%
18
BMC Genomics
○ 406 papers in training set
Top 5%
1.2%
19
Bioinformatics
◐ 1204 papers in training set
Top 7%
1.2%
20
PeerJ
◐ 308 papers in training set
Top 9%
1.0%
21
NeuroImage
○ 903 papers in training set
Top 6%
1.0%
22
The Journals of Gerontology, Series A: Biological Sciences and Medical Sciences
◐ 26 papers in training set
Top 0.5%
0.9%
23
Genome Biology and Evolution
◐ 338 papers in training set
Top 3%
0.9%
24
Computational and Structural Biotechnology Journal
● 242 papers in training set
Top 6%
0.9%
25
PLOS Genetics
● 862 papers in training set
Top 11%
0.9%
26
Nucleic Acids Research
◐ 1281 papers in training set
Top 14%
0.6%
27
Genetic Epidemiology
○ 55 papers in training set
Top 0.7%
0.6%
28
Neurobiology of Aging
○ 107 papers in training set
Top 2%
0.6%
29
Genome Research
● 468 papers in training set
Top 8%
0.5%
30
Patterns
○ 78 papers in training set
Top 3%
0.5%

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.