Source Data for "Multimodal Prediction of Primary Open-Angle Glaucoma Using Polygenic Risk Scores and Clinical Features in a High-Risk African Ancestry Cohort" (iScience, 2026)
Description
Source data for the revised manuscript "Multimodal Prediction of Primary Open-Angle Glaucoma Using Polygenic Risk Scores and Clinical Features in a High-Risk African Ancestry Cohort" (iScience, ISCIENCE-D-26-03991). Version 2 (September 2026) replaces Version 1 (May 2026), whose results predate the corrections described in the revised manuscript (PGS616 transformation, PMBB age restriction, and participant-level bootstrap inference). Contents: Supplemental Tables S1-S20 as submitted; the result tables written by the analysis code (GitHub yzhu13/poag-pgs-ml, release v3-revision), with repeated cross-validation given as mean and SD only; and the per-replicate out-of-bag AUCs of the participant-level bootstrap (B = 2,000), from which every training-cohort interval and paired increment can be recomputed. README.txt lists the files. No individual-level data are included; POAAGG and Penn Medicine BioBank data are available under controlled access (dbGaP phs001312 and phs001453).
Files
Steps to reproduce
The tables were produced by the analysis code in GitHub repository yzhu13/poag-pgs-ml, release v3-revision (https://github.com/yzhu13/poag-pgs-ml/releases/tag/v3-revision), run on the controlled-access POAAGG and Penn Medicine BioBank data (dbGaP phs001312, phs001453) in the order given in the repository README, with the package versions pinned in requirements.txt (Python 3.13.7, scikit-learn 1.7.2). Training-cohort confidence intervals and paired increments can be recomputed without refitting from the bootstrap replicate files deposited here (script 17 in the repository redraws the bootstrap figure panels from them).
Institutions
- University of PennsylvaniaPennsylvania, Philadelphia