5 ms·
I was the person at Google who originally set them on the path of mirroring large amounts of Genomics data in Google Cloud. The SRA was one example- the NIH se
by dekhn 5y ago
I was the person at Google who originally set them on the path of mirroring large amounts of Genomics data in Google Cloud. The SRA was one example- the NIH serves it up on a very clunky and slow (no good for high throughput workflows) so many people, including Google, make a controlled-access mirror in GCS/S3/Blobstore.
In all cases, the mirror should reflect what's served on SRA, which means that if there's deletions, GCS should delete it as well (as soon as technically possible). It's unclear whether Google set up the infra to delete items from buckets (they should have, but it was probably a low priority).
I wonder what other deleted sequences can be recovered from the bucket. I hope the author told the Google Cloud admins about their problem.