GoKawiilPsychAD study builds cell-type gene expression models from over 6 million brain nuclei
Researchers from the PsychAD consortium describe training single-nucleus transcriptomic imputation models (snTIMs) using single-nucleus RNA sequencing data from more than 6 million nuclei drawn from brain tissue of 1,494 donors. The work involved dividing donors into ancestry groups, harmonizing genotype data across multiple sequencing and array platforms, and building ancestry-specific reference panels to align training and target cohorts, including data from the Million Veteran Program.
GoKawiil's interpretation of the reporting above, not reported fact.
This kind of large-scale, ancestry-aware modeling infrastructure could improve the accuracy of transcriptome-wide association studies for brain disorders by capturing cell-type-specific gene expression patterns rather than relying on bulk tissue averages. Because the researchers explicitly excluded some ancestry groups for lack of statistical power, the resulting models may work better for populations of European, African and admixed American ancestry than for East Asian or South Asian populations, a limitation that could affect how broadly these tools generalize.
- The PsychAD dataset includes snRNA-seq data from over 6 million nuclei across 1,494 donors' brain tissue.
- Ancestry-specific reference panels were built to harmonize genotype data across multiple sequencing platforms, including for the Million Veteran Program.
- East Asian and South Asian ancestry groups were excluded from model training due to insufficient sample sizes.
Source: nature.com — Venkatesh, 2026-09-23
Published there as: “Single-nucleus transcriptome-wide association study of human brain disorders”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.