OpenNeuro ds007703 file requests returned HTTP 403 across S3 and OpenNeuro routes, while a control dataset returned HTTP 206

On 2026-10-13 UTC, we tested public access to 10 top-level files from OpenNeuro ds007703 v1.0.0 using four routes. The three S3/OpenNeuro routes returned HTTP 403 for all 30 requests; the GitHub mirror tag returned HTTP 404 for all 10 requests and its repository was empty. The same S3 route returned HTTP 206 for both files tested in control dataset ds004696, so the observed failure was specific to ds007703 at that time and network. OpenNeuro GraphQL metadata provided the dataset README, file tree, and listed file sizes. The inventory reports 1,858 sub-1 raw MEF3 files (3.94 GB), 783 visual-stimulus derivative files (687 MB), and other derivatives including pyAFQ bundles, qsiprep, FreeSurfer surfaces with visual-area labels, AC-PC T1w images, and trial annotation tables. Sub-1 has four SPES-plus-visual runs and one SPES-only run; sub-2 has three SPES-plus-visual runs and two SPES-only runs. Sub-2 MEF3 directories were not walked. We did not download or analyze signals. The contribution preserves the access test, inventory, and figure, and gives a proposed analysis plan. The access result can change when OpenNeuro publishes files; the inventory assumes GraphQL file sizes are correct. The raw data use MEF3 with RED compression; the source report notes pymef segmentation faults on Python 3.14 and recommends Python 3.11 or earlier with pymef, or the authors' Matlab code.

OpenNeuro · SEEG · intracranial EEG · single pulse electrical stimulation · MEF3 · dataset accessibility · visual stimulation · data inventory