{
  "id": 577781,
  "title": "Any recommendation on long RNA sequence using RibonanzaNet.",
  "url": "/competitions/stanford-rna-3d-folding/discussion/577781",
  "author_name": "",
  "post_date": "2025-05-07T04:26:53.784917300Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi, <br>\nI'm trying to <strong>predict secondary structure of \"train_sequences.csv' sequences using RibonanzaNet.</strong><br>\neven though I'm using RunPod with <strong>2 x RTX A6000(total 96GB)</strong> <br>\nwith accelerator library (accelerator launch infer.py), im keep getting CUDA out of memory issue.</p>\n<p>my question is<br>\n1) any suggestion on <strong>inferring secondary structure of long sequences</strong> such as [1513, 2686, 2685, 2884] length RNA sequences?</p>\n<p>2) any suggestion on <strong>getting secondary structure of whole sequences</strong> other than ribonanzanet?</p>\n<p>3) any suggestion on <strong>handling RNA sequences that contain 'X' or '-'</strong> ?</p>",
  "messages": [
    {
      "id": "3196466",
      "postDate": "05/07/2025 04:26:53",
      "content": "<p>Hi, <br>\nI'm trying to <strong>predict secondary structure of \"train_sequences.csv' sequences using RibonanzaNet.</strong><br>\neven though I'm using RunPod with <strong>2 x RTX A6000(total 96GB)</strong> <br>\nwith accelerator library (accelerator launch infer.py), im keep getting CUDA out of memory issue.</p>\n<p>my question is<br>\n1) any suggestion on <strong>inferring secondary structure of long sequences</strong> such as [1513, 2686, 2685, 2884] length RNA sequences?</p>\n<p>2) any suggestion on <strong>getting secondary structure of whole sequences</strong> other than ribonanzanet?</p>\n<p>3) any suggestion on <strong>handling RNA sequences that contain 'X' or '-'</strong> ?</p>",
      "rawMarkdown": "Hi, \nI'm trying to **predict secondary structure of \"train_sequences.csv' sequences using RibonanzaNet.**\neven though I'm using RunPod with **2 x RTX A6000(total 96GB)** \nwith accelerator library (accelerator launch infer.py), im keep getting CUDA out of memory issue.\n\nmy question is\n1) any suggestion on **inferring secondary structure of long sequences** such as [1513, 2686, 2685, 2884] length RNA sequences?\n\n2) any suggestion on **getting secondary structure of whole sequences** other than ribonanzanet?\n\n3) any suggestion on **handling RNA sequences that contain 'X' or '-'** ?",
      "votes": null
    },
    {
      "id": "3196570",
      "postDate": "05/07/2025 07:05:31",
      "content": "<p>You may want to try this: <a href=\"https://github.com/febos/SQUARNA\" target=\"_blank\">https://github.com/febos/SQUARNA</a><br>\nIt handles gaps (like '-') and ignores non-ACGU symbols (like 'X'), and it has specific configs for long sequences. Although you should not expect accurate predictions for &gt; 1000nt sequences from any tool.</p>",
      "rawMarkdown": "You may want to try this: https://github.com/febos/SQUARNA\nIt handles gaps (like '-') and ignores non-ACGU symbols (like 'X'), and it has specific configs for long sequences. Although you should not expect accurate predictions for > 1000nt sequences from any tool.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3196570,
      "author_name": "eugenebaulin",
      "author_url": "",
      "post_date": "05/07/2025 07:05:31",
      "content": "<p>You may want to try this: <a href=\"https://github.com/febos/SQUARNA\" target=\"_blank\">https://github.com/febos/SQUARNA</a><br>\nIt handles gaps (like '-') and ignores non-ACGU symbols (like 'X'), and it has specific configs for long sequences. Although you should not expect accurate predictions for &gt; 1000nt sequences from any tool.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3196466": "Hi, \nI'm trying to **predict secondary structure of \"train_sequences.csv' sequences using RibonanzaNet.**\neven though I'm using RunPod with **2 x RTX A6000(total 96GB)** \nwith accelerator library (accelerator launch infer.py), im keep getting CUDA out of memory issue.\n\nmy question is\n1) any suggestion on **inferring secondary structure of long sequences** such as [1513, 2686, 2685, 2884] length RNA sequences?\n\n2) any suggestion on **getting secondary structure of whole sequences** other than ribonanzanet?\n\n3) any suggestion on **handling RNA sequences that contain 'X' or '-'** ?",
    "3196570": "You may want to try this: https://github.com/febos/SQUARNA\nIt handles gaps (like '-') and ignores non-ACGU symbols (like 'X'), and it has specific configs for long sequences. Although you should not expect accurate predictions for > 1000nt sequences from any tool."
  },
  "source": "meta"
}