Visual Learning Ophys: z-drift QC for early TRAINING_3 sessions, and public release of the 2026-09-19 coreg-id-mapping generation

Technical
1

I’m analyzing the pre-release Visual Learning Ophys dataset (public data in s3://aind-open-data/) and have three questions that the Data Book and public repos don’t answer. I’ve also emailed aindopendata@alleninstitute.org.

1. Z-drift QC backfill. The TRAINING_3_images_A_10uL_reward sessions below have no “Z-drift Analysis” evaluation in quality_control.json, apparently because their processing generation predates the z-drift evaluator (8 of the 19 TRAINING_3 sessions):

  • 800792: 2025-08-04, 08-05, 08-07, 08-08
  • 800995: 2025-08-14, 08-15
  • 804363: 2025-08-18, 08-20

Is reprocessing or z-drift evaluation planned for these sessions?

2. 782149, OPHYS_4_images_B on 2025-04-28 (multiplane-ophys_782149_2025-04-28_10-02-28). The public 2026-07-06 coreg-id-mapping table for 782149 has no rows for this session. Does the 2026-09-19 coreg-id-mapping generation (currently in the Code Ocean bucket) include it?

3. Is the 2026-09-19 coreg-id-mapping generation scheduled for transfer to s3://aind-open-data/?

Related, already reported:

Thanks!

2

I am looking into these this week. Thanks for reaching out!

3

Thank you!

4

Hi @Archishman770, thanks for the detailed questions. As you noted, Visual Learning is still pre-release. We’ve made it available for exploratory use, but it’s still being curated, so please treat any results as preliminary.

1. Z-drift QC: @seanmcculloch is working on the backfill for those sessions, and we’ll post here once the results are final.

2 & 3. Coreg-ID mapping: Please use this newer asset, which is now on the public bucket:

s3://aind-open-data/cell-types-and-learning-data/consensus-clustering/hcr-inhibitory-consensus-clusters_2026-10-26_01-51-27/
(Code Ocean asset ID: 8b915674-2cb3-48a9-b770-709c83616482)

Its coreg_mapping_table.csv includes 782149’s 2025-04-28 OPHYS_4 session, and subject IDs are consistent across all rows. The asset also contains an HCR cell × gene table for inhibitory neurons across all 6 mice, plus extras such as a mapping to the Patch-seq dataset for validation.

Thanks for flagging these issues; it really helps us clean things up before publication. Feel free to reach out at marinag@alleninstitute.org if you’d like to discuss your analysis in more detail. We’re always glad to hear how people are using the data.

-Marina