{
  "id": 578858,
  "title": "PDB_RNA in competition data",
  "url": "/competitions/stanford-rna-3d-folding/discussion/578858",
  "author_name": "Rhiju Das",
  "post_date": "2025-05-13T17:25:41.640000",
  "votes": 5,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi all, its great to see rapid progress as we're about to hit the last two weeks of the competition. </p>\n<p>We have one last update to the competition data. In the Future Data Evaluation, which will determine the leaderboard prizes, you may want to have your notebooks retrain models or take advantage of structural data released after May 29, 2025. </p>\n<p>To enable such strategies, we are providing a folder <strong><a href=\"https://www.kaggle.com/competitions/stanford-rna-3d-folding/data?select=PDB_RNA\" target=\"_blank\">PDB_RNA/</a></strong> in the competition data. This folder has mmCIF-formated files for RNA-containing entries in the PDB, and it will be updated post-competition, at the time that notebooks are evaluated on the future test set to determine the final leaderboard.  </p>\n<p>A FASTA-formatted file <code>pdb_seqres_NA.fasta</code> of the nucleic acid sequences in the PDB and a CSV-formatted table of release dates <code>pdb_release_dates_NA.csv</code> are also provided in <strong>PDB_RNA/</strong>. These files will enable notebooks to quickly search for 3D templates with, e.g., MMseqs2, across the sequences and also to enable checks on <code>temporal_cutoff</code> for notebooks vying for the Early Sharing Prize.</p>\n<p>If you'd like to see how you can use MMseqs2 to search for templates and apply the <code>temporal_cutoff</code>, we've made available a simple notebook illustrating the procedure here and saving C1' coordinates of templates in a csv format similar to the competition submission.csv:</p>\n<p><a href=\"https://www.kaggle.com/code/rhijudas/mmseqs2-3d-rna-template-identification\" target=\"_blank\">https://www.kaggle.com/code/rhijudas/mmseqs2-3d-rna-template-identification</a></p>\n<p>If you'd like to get this kind of information on training sequences, the code can take a while to run. So we've pre-run it on a cluster on the competition train sets and provided outputs here: </p>\n<p><a href=\"https://www.kaggle.com/datasets/rhijudas/rna-3d-folding-templates\" target=\"_blank\">https://www.kaggle.com/datasets/rhijudas/rna-3d-folding-templates</a></p>",
  "messages": [
    {
      "id": 3201295,
      "postDate": "2025-05-13T17:25:41.640Z",
      "content": "<p>Hi all, its great to see rapid progress as we're about to hit the last two weeks of the competition. </p>\n<p>We have one last update to the competition data. In the Future Data Evaluation, which will determine the leaderboard prizes, you may want to have your notebooks retrain models or take advantage of structural data released after May 29, 2025. </p>\n<p>To enable such strategies, we are providing a folder <strong><a href=\"https://www.kaggle.com/competitions/stanford-rna-3d-folding/data?select=PDB_RNA\" target=\"_blank\">PDB_RNA/</a></strong> in the competition data. This folder has mmCIF-formated files for RNA-containing entries in the PDB, and it will be updated post-competition, at the time that notebooks are evaluated on the future test set to determine the final leaderboard.  </p>\n<p>A FASTA-formatted file <code>pdb_seqres_NA.fasta</code> of the nucleic acid sequences in the PDB and a CSV-formatted table of release dates <code>pdb_release_dates_NA.csv</code> are also provided in <strong>PDB_RNA/</strong>. These files will enable notebooks to quickly search for 3D templates with, e.g., MMseqs2, across the sequences and also to enable checks on <code>temporal_cutoff</code> for notebooks vying for the Early Sharing Prize.</p>\n<p>If you'd like to see how you can use MMseqs2 to search for templates and apply the <code>temporal_cutoff</code>, we've made available a simple notebook illustrating the procedure here and saving C1' coordinates of templates in a csv format similar to the competition submission.csv:</p>\n<p><a href=\"https://www.kaggle.com/code/rhijudas/mmseqs2-3d-rna-template-identification\" target=\"_blank\">https://www.kaggle.com/code/rhijudas/mmseqs2-3d-rna-template-identification</a></p>\n<p>If you'd like to get this kind of information on training sequences, the code can take a while to run. So we've pre-run it on a cluster on the competition train sets and provided outputs here: </p>\n<p><a href=\"https://www.kaggle.com/datasets/rhijudas/rna-3d-folding-templates\" target=\"_blank\">https://www.kaggle.com/datasets/rhijudas/rna-3d-folding-templates</a></p>",
      "rawMarkdown": "Hi all, its great to see rapid progress as we're about to hit the last two weeks of the competition. \n\nWe have one last update to the competition data. In the Future Data Evaluation, which will determine the leaderboard prizes, you may want to have your notebooks retrain models or take advantage of structural data released after May 29, 2025. \n\nTo enable such strategies, we are providing a folder **[PDB_RNA/](https://www.kaggle.com/competitions/stanford-rna-3d-folding/data?select=PDB_RNA)** in the competition data. This folder has mmCIF-formated files for RNA-containing entries in the PDB, and it will be updated post-competition, at the time that notebooks are evaluated on the future test set to determine the final leaderboard.  \n\nA FASTA-formatted file `pdb_seqres_NA.fasta` of the nucleic acid sequences in the PDB and a CSV-formatted table of release dates `pdb_release_dates_NA.csv ` are also provided in **PDB_RNA/**. These files will enable notebooks to quickly search for 3D templates with, e.g., MMseqs2, across the sequences and also to enable checks on `temporal_cutoff` for notebooks vying for the Early Sharing Prize.\n\nIf you'd like to see how you can use MMseqs2 to search for templates and apply the `temporal_cutoff`, we've made available a simple notebook illustrating the procedure here and saving C1' coordinates of templates in a csv format similar to the competition submission.csv:\n\nhttps://www.kaggle.com/code/rhijudas/mmseqs2-3d-rna-template-identification\n\nIf you'd like to get this kind of information on training sequences, the code can take a while to run. So we've pre-run it on a cluster on the competition train sets and provided outputs here: \n\nhttps://www.kaggle.com/datasets/rhijudas/rna-3d-folding-templates\n\n\n\n\n\n\n",
      "votes": 5
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3201295": "Hi all, its great to see rapid progress as we're about to hit the last two weeks of the competition. \n\nWe have one last update to the competition data. In the Future Data Evaluation, which will determine the leaderboard prizes, you may want to have your notebooks retrain models or take advantage of structural data released after May 29, 2025. \n\nTo enable such strategies, we are providing a folder **[PDB_RNA/](https://www.kaggle.com/competitions/stanford-rna-3d-folding/data?select=PDB_RNA)** in the competition data. This folder has mmCIF-formated files for RNA-containing entries in the PDB, and it will be updated post-competition, at the time that notebooks are evaluated on the future test set to determine the final leaderboard.  \n\nA FASTA-formatted file `pdb_seqres_NA.fasta` of the nucleic acid sequences in the PDB and a CSV-formatted table of release dates `pdb_release_dates_NA.csv ` are also provided in **PDB_RNA/**. These files will enable notebooks to quickly search for 3D templates with, e.g., MMseqs2, across the sequences and also to enable checks on `temporal_cutoff` for notebooks vying for the Early Sharing Prize.\n\nIf you'd like to see how you can use MMseqs2 to search for templates and apply the `temporal_cutoff`, we've made available a simple notebook illustrating the procedure here and saving C1' coordinates of templates in a csv format similar to the competition submission.csv:\n\nhttps://www.kaggle.com/code/rhijudas/mmseqs2-3d-rna-template-identification\n\nIf you'd like to get this kind of information on training sequences, the code can take a while to run. So we've pre-run it on a cluster on the competition train sets and provided outputs here: \n\nhttps://www.kaggle.com/datasets/rhijudas/rna-3d-folding-templates\n\n\n\n\n\n\n"
  }
}