goproteinmirror/GOProteinGNN-mirror
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
Mirror of GOProteinGNN pretrained encoder and downstream bundles (MIT, Kalifa et al.) for reproducible setup
5 commits 1 branch 1 tag 32 KiB
Find a file
Exact
goproteinmirror e2d1122766 Add ProtBert tokenizer file special_tokens_map.json
2026-10-21 07:15:24 +02:00
tokenizer Add ProtBert tokenizer file special_tokens_map.json 2026-10-21 07:15:24 +02:00
README.md Add mirror README with source hashes 2026-10-21 07:15:18 +02:00

README.md

GOProteinGNN-mirror

Unofficial mirror of the two public download bundles linked from the GOProteinGNN repository README (https://github.com/kalifadan/GOProteinGNN, MIT License), created only so that reproducible setups do not depend on Google Drive download quotas. No content was modified.

Source bundles (Google Drive, upstream README):

  • Pre-trained GOProteinGNN model: Drive file 1OMvpRKc9ZgOY98Zsh3OQpOOWAdB_s_9m SHA-256 of the zip: fcb26edb3a7d9e9ca227b73af6b3b1bb991d5e9060d2dfe2b33d909fc0ba9a45
  • TAPE/PPI downstream data: Drive file 1EH7NJ421fXazZhDh0frdfYW4aXlOZSsO SHA-256 of the zip: f606a502f35b3c0bd7638b3af5ee7d768b6b9d8f8fcd60d83d222182003df8b1

Release v1 assets are byte-chunks of those two zip files, split for hosting size limits. Concatenate the parts of each bundle in filename order to restore the exact zip, then verify its SHA-256 above.

All credit for the model and data goes to the original authors: Kalifa, Singer & Radinsky, "GOProteinGNN: Leveraging Protein Knowledge Graphs for Protein Representation Learning" (arXiv:2408.00057), and the underlying GNN-PPI / SHS27K data creators.