Posted By: The American Journal of Human Genetics, AJHG
Each month, the editors of The American Journal of Human Genetics interview an author of a recently published paper. This month, we check in with Xinyu to discuss his recent paper, “Multi-ancestry transcriptome-wide association study reveals shared and population-specific genetic effects in Alzheimer’s disease.”
AJHG: What motivated you to start working on this project?
XS: Alzheimer’s disease risk differs across populations, but much of the genetic and transcriptomic work in the field has historically focused on cohorts of European ancestry. That makes it harder to determine which disease-associated regulatory mechanisms are shared across populations and which may be population-specific.
This project grew out of that gap. We had access to MAGENTA, which includes whole-blood RNA-seq and genotype data from African American, Hispanic, and non-Hispanic White participants, and we wanted to use that resource to study Alzheimer’s disease through a more population-aware transcriptomic framework. SuShiE gave us a way to fine-map cis-eQTL effects jointly across populations and then build ancestry-matched TWAS models, so we could ask not only which genes were associated with Alzheimer’s disease, but also which regulatory variants were driving those signals.
AJHG: What about the paper/project most excites you?
XS: What excites me most is that the project makes TWAS results more interpretable. TWAS can identify genes whose genetically predicted expression is associated with disease, but it is often difficult to tell which regulatory variants are responsible for the signal. By combining multi-population eQTL fine-mapping with TWAS, we were able to narrow many associations to compact credible sets and compare those fine-mapped eQTL variants with known GWAS signals.
For established Alzheimer’s disease loci such as BIN1, PTK2B, and DMPK, this helped refine the regulatory interpretation beyond the sentinel GWAS variants. We also identified COG4 as a candidate Alzheimer’s disease gene in non-Hispanic White participants, with functional evidence pointing toward distal enhancer-mediated regulation.
AJHG: Thinking about the bigger picture, what implications do you see from this work for the larger human genetics community?
XS: One broader implication is that diverse datasets are scientifically important, not only necessary for equity. Differences in linkage disequilibrium and allele frequency across populations can help refine association signals and improve fine-mapping resolution. In our study, including all three populations reduced the median number of variants per credible set and helped identify both shared and population-specific regulatory patterns.
At the same time, the work also points to a major limitation in the field: current GWAS and molecular QTL resources are still much better powered in European-ancestry populations than in many other groups. For human genetics to produce findings that are generalizable and mechanistically useful, we need larger and better-balanced multi-ancestry GWAS, eQTL, and tissue-specific datasets, along with methods that can model both shared and population-specific genetic effects.
AJHG: What advice do you have for trainees/young scientists?
XS: I am still a trainee myself, so my advice is really what I try to practice: learn enough across disciplines to question your own results from multiple angles. In computational genetics, it is easy to treat a method as a black box, especially when the pipeline is complex. But the most useful insights often come from simple, careful questions: Does this result make biological sense? Could it be driven by linkage disequilibrium, sample size, tissue context, or model assumptions? What would convince me that this signal is real?
I would also encourage trainees to seek collaborators who think differently from them. This project required genetics, statistics, computation, and Alzheimer’s disease biology. Being able to communicate across those areas is just as important as technical skill.
AJHG: And for fun, tell us something about your life outside of the lab.
XS: Outside the lab, I am pretty low-key. I like spending weekends watching good movies, working on programming side projects, and tinkering with homelab setups. I enjoy the mix of storytelling, problem-solving, and the occasional satisfaction of getting a stubborn system to finally work.
Xinyu Sun is a PhD candidate in Biomedical & Health Informatics in the Department of Population and Quantitative Health Sciences, School of Medicine, Case Western Reserve University.