Congratulations to everyone who participated in the Stanford RNA 3D Folding challenge!
\nWe have passed the Final Submission Deadline, with numerous teams outperforming the human expert baseline. But the competition is not over.
\nTo ensure rigor, we will be evaluating notebooks over the next 3-4 months based on ‘future’ RNA targets whose 3D structures are released between now and September.
\nWe plan to score and release the final Private Leaderboard by end of September, 2025. At that time, we will announce leaderboard prize winners. And we will report whether or not we’ve collectively made a major advance on one of biology’s remaining grand challenges!
\nIn the meanwhile, hosts and scientific collaborators will be working to synthesize insights from the current competition to develop even better models. And we need your help!
\nHere are some things you can do to help us further push the envelope on RNA 3D structure prediction:
\nvfold_human_expert baseline, how did you do it? Is your notebook eligible for the Early Sharing Prize?For the hosts,
\nRhiju
Congratulations to everyone who participated in the Stanford RNA 3D Folding challenge!
\nWe have passed the Final Submission Deadline, with numerous teams outperforming the human expert baseline. But the competition is not over.
\nTo ensure rigor, we will be evaluating notebooks over the next 3-4 months based on ‘future’ RNA targets whose 3D structures are released between now and September.
\nWe plan to score and release the final Private Leaderboard by end of September, 2025. At that time, we will announce leaderboard prize winners. And we will report whether or not we’ve collectively made a major advance on one of biology’s remaining grand challenges!
\nIn the meanwhile, hosts and scientific collaborators will be working to synthesize insights from the current competition to develop even better models. And we need your help!
\nHere are some things you can do to help us further push the envelope on RNA 3D structure prediction:
\nvfold_human_expert baseline, how did you do it? Is your notebook eligible for the Early Sharing Prize?For the hosts,
\nRhiju
Thank you for hosting the great competition!
\nAs I posted https://www.kaggle.com/competitions/stanford-rna-3d-folding/discussion/582295 , my approach was mainly TBM with very recent DB.
\nI used blastn search with Smith-Waterman alignment + representation based sequence search & alignment (like PLMsearch https://www.nature.com/articles/s41467-024-46808-5) using ribonanzanet representation.
\nMissing regions were filled by Boltz-1 or Chai-1 prediction or other templates.
\nGenerally, I used cosine similarity or Pearson correlation coeffitient for scoring function of representation based sequence search & alignment and it improved my score approximately 0.14 compared to blastn search (70% confidence. There were complicated issues).
\nMy last submission (0.671) used dot product for scoring function + fragment masking (gappy short alignment regions were masked) and it improved my score around 0.07. I don't know which factor was critical.
\nIn my previous experience, dot product was not so good thus I submitted it just for trial.
\nUpdate (2025/06/17)
\nI uploaded my presentation slide in my discussion page.
\nPlease feel free to post comments if you have any questions.
\nhttps://www.kaggle.com/competitions/stanford-rna-3d-folding/discussion/582295
Can you clarify? I wasn't able to form a submission by the first deadline but I have a solution that I am sure will make the leaderboard I just have to complete work on it. Am I still able to make a final submission and compete for the September deadline?
", "rawMarkdown": "Can you clarify? I wasn't able to form a submission by the first deadline but I have a solution that I am sure will make the leaderboard I just have to complete work on it. Am I still able to make a final submission and compete for the September deadline?", "replies": [ { "id": 3248251, "postDate": "2025-07-14T07:58:06.470Z", "content": "No submissions can be made after the first deadline, @tuttlen
", "rawMarkdown": "No submissions can be made after the first deadline, @tuttlen " } ] }, { "id": 3235093, "postDate": "2025-06-28T17:51:41.487Z", "content": "@rhijudas Are you going to publish interim results in near future?
", "rawMarkdown": "@rhijudas Are you going to publish interim results in near future?" }, { "id": 3218228, "postDate": "2025-06-06T02:12:45.430Z", "content": "Hi Rhiju,
\nThank you for this incredible opportunity to continue testing and refining our solutions during this extended evaluation period!
\nI'm really excited about the chance to further develop and validate the approach I've been working on. This additional time is invaluable for pushing the boundaries of what's possible in RNA structure prediction.
@rhijudas If we discover an interesting approach that performs better than the current public leaderboard, will it count toward the final winning list?
", "rawMarkdown": "@rhijudas If we discover an interesting approach that performs better than the current public leaderboard, will it count toward the final winning list?", "replies": [ { "id": 3213842, "postDate": "2025-05-30T14:47:35.043Z", "content": "As with all Kaggle competitions, any new submissions after the deadline will not show up on the leaderboard. However, if you post your public LB score & link to notebook here on Discussion, hosts will certainly pay attention. If your innovation can be integrated into an improved final model, it will be described in the scientific paper from this competition.
", "rawMarkdown": "As with all Kaggle competitions, any new submissions after the deadline will not show up on the leaderboard. However, if you post your public LB score & link to notebook here on Discussion, hosts will certainly pay attention. If your innovation can be integrated into an improved final model, it will be described in the scientific paper from this competition." } ] }, { "id": 3213796, "postDate": "2025-05-30T13:24:31.310Z", "content": "However, the competition information now shows: \"Submissions currently disabled,\" which means that from now on, it is no longer possible to make submissions. So regarding what you mentioned: \"If you get new ideas now after Final Submission Deadline, remember that you can still submit notebooks for scoring. Your scores won’t show up on leaderboards but if you find something interesting, please post in the Discussion.\"—how can this be done?
", "rawMarkdown": "However, the competition information now shows: \"Submissions currently disabled,\" which means that from now on, it is no longer possible to make submissions. So regarding what you mentioned: \"If you get new ideas now after Final Submission Deadline, remember that you can still submit notebooks for scoring. Your scores won’t show up on leaderboards but if you find something interesting, please post in the Discussion.\"—how can this be done?", "replies": [ { "id": 3213816, "postDate": "2025-05-30T14:00:35.240Z", "rawMarkdown": "", "isDeleted": true }, { "id": 3213845, "postDate": "2025-05-30T14:49:55.703Z", "content": "Thanks for pointing this out. There is some behind-the-scenes cleanup happening related to the Final Submission Deadline, and we are temporarily disabling submissions, but you'll see submissions re-enabled after that. I'm editing the post above to explain.
", "rawMarkdown": "Thanks for pointing this out. There is some behind-the-scenes cleanup happening related to the Final Submission Deadline, and we are temporarily disabling submissions, but you'll see submissions re-enabled after that. I'm editing the post above to explain.", "replies": [ { "id": 3255559, "postDate": "2025-07-29T00:44:47.667Z", "content": "is there an update on this now?
", "rawMarkdown": "is there an update on this now?" } ] } ] }, { "id": 3213821, "postDate": "2025-05-30T14:09:01.917Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 3214427, "author_name": "odat", "author_url": "", "post_date": "2025-05-31T14:14:18.387000", "content": "Thank you for hosting the great competition!
\nAs I posted https://www.kaggle.com/competitions/stanford-rna-3d-folding/discussion/582295 , my approach was mainly TBM with very recent DB.
\nI used blastn search with Smith-Waterman alignment + representation based sequence search & alignment (like PLMsearch https://www.nature.com/articles/s41467-024-46808-5) using ribonanzanet representation.
\nMissing regions were filled by Boltz-1 or Chai-1 prediction or other templates.
\nGenerally, I used cosine similarity or Pearson correlation coeffitient for scoring function of representation based sequence search & alignment and it improved my score approximately 0.14 compared to blastn search (70% confidence. There were complicated issues).
\nMy last submission (0.671) used dot product for scoring function + fragment masking (gappy short alignment regions were masked) and it improved my score around 0.07. I don't know which factor was critical.
\nIn my previous experience, dot product was not so good thus I submitted it just for trial.
\nUpdate (2025/06/17)
\nI uploaded my presentation slide in my discussion page.
\nPlease feel free to post comments if you have any questions.
\nhttps://www.kaggle.com/competitions/stanford-rna-3d-folding/discussion/582295
Can you clarify? I wasn't able to form a submission by the first deadline but I have a solution that I am sure will make the leaderboard I just have to complete work on it. Am I still able to make a final submission and compete for the September deadline?
", "votes": 0, "replies": [ { "id": 3248251, "author_name": "Arunodhayan", "author_url": "", "post_date": "2025-07-14T07:58:06.470000", "content": "No submissions can be made after the first deadline, @tuttlen
", "votes": 0, "replies": [] } ] }, { "id": 3235093, "author_name": "Ogurtsov", "author_url": "", "post_date": "2025-06-28T17:51:41.487000", "content": "@rhijudas Are you going to publish interim results in near future?
", "votes": 0, "replies": [] }, { "id": 3218228, "author_name": "Fernando Rodrigues", "author_url": "", "post_date": "2025-06-06T02:12:45.430000", "content": "Hi Rhiju,
\nThank you for this incredible opportunity to continue testing and refining our solutions during this extended evaluation period!
\nI'm really excited about the chance to further develop and validate the approach I've been working on. This additional time is invaluable for pushing the boundaries of what's possible in RNA structure prediction.
@rhijudas If we discover an interesting approach that performs better than the current public leaderboard, will it count toward the final winning list?
", "votes": 0, "replies": [ { "id": 3213842, "author_name": "Rhiju Das", "author_url": "", "post_date": "2025-05-30T14:47:35.043000", "content": "As with all Kaggle competitions, any new submissions after the deadline will not show up on the leaderboard. However, if you post your public LB score & link to notebook here on Discussion, hosts will certainly pay attention. If your innovation can be integrated into an improved final model, it will be described in the scientific paper from this competition.
", "votes": 0, "replies": [] } ] }, { "id": 3213796, "author_name": "Ray", "author_url": "", "post_date": "2025-05-30T13:24:31.310000", "content": "However, the competition information now shows: \"Submissions currently disabled,\" which means that from now on, it is no longer possible to make submissions. So regarding what you mentioned: \"If you get new ideas now after Final Submission Deadline, remember that you can still submit notebooks for scoring. Your scores won’t show up on leaderboards but if you find something interesting, please post in the Discussion.\"—how can this be done?
", "votes": 0, "replies": [ { "id": 3213816, "author_name": "", "author_url": "", "post_date": "2025-05-30T14:00:35.240000", "content": "", "votes": 0, "replies": [] }, { "id": 3213845, "author_name": "Rhiju Das", "author_url": "", "post_date": "2025-05-30T14:49:55.703000", "content": "Thanks for pointing this out. There is some behind-the-scenes cleanup happening related to the Final Submission Deadline, and we are temporarily disabling submissions, but you'll see submissions re-enabled after that. I'm editing the post above to explain.
", "votes": 0, "replies": [ { "id": 3255559, "author_name": "hari kishan123", "author_url": "", "post_date": "2025-07-29T00:44:47.667000", "content": "is there an update on this now?
", "votes": 0, "replies": [] } ] } ] }, { "id": 3213821, "author_name": "", "author_url": "", "post_date": "2025-05-30T14:09:01.917000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "3213787": "Congratulations to everyone who participated in the Stanford RNA 3D Folding challenge! \n\nWe have passed the Final Submission Deadline, with numerous teams outperforming the human expert baseline. But the competition is not over. \n\nTo ensure rigor, we will be evaluating notebooks over the next 3-4 months based on ‘future’ RNA targets whose 3D structures are released between now and September. \n\nWe plan to score and release the final Private Leaderboard by end of September, 2025. At that time, we will announce leaderboard prize winners. And we will report whether or not we’ve collectively made a major advance on one of biology’s remaining grand challenges!\n\nIn the meanwhile, hosts and scientific collaborators will be working to synthesize insights from the current competition to develop even better models. And we need your help! \n\nHere are some things you can do to help us further push the envelope on RNA 3D structure prediction:\n\n- While the competition is fresh in your minds, make a new post here in the Discussion of what you tried — what worked and what didn’t work?\n- Consider making your notebook public, and linking in your post!\n- For the teams who beat the `vfold_human_expert` baseline, how did you do it? Is your notebook eligible for the Early Sharing Prize?\n- If you get new ideas now after Final Submission Deadline, we will soon reactivate submissions so you can submit future notebooks for scoring. Your post-deadline submission scores won’t show up on leaderboards but if you find something interesting, please post your score and notebook in the Discussion.\n- Can you think of new experimental data sources that might improve your model? For example, are there particular sets of RNA molecules on which we could carry out chemical mapping in our efforts to scale up [RibonanzaNet](https://www.kaggle.com/competitions/stanford-rna-3d-folding/discussion/565306)?\n- Keep notifications on for this competition’s Discussion — we anticipate some exciting updates through the summer!\n\nFor the hosts,\nRhiju", "3214427": "Thank you for hosting the great competition!\n\nAs I posted https://www.kaggle.com/competitions/stanford-rna-3d-folding/discussion/582295 , my approach was mainly TBM with very recent DB.\n\nI used blastn search with Smith-Waterman alignment + representation based sequence search & alignment (like PLMsearch https://www.nature.com/articles/s41467-024-46808-5) using ribonanzanet representation.\n\nMissing regions were filled by Boltz-1 or Chai-1 prediction or other templates.\n\nGenerally, I used cosine similarity or Pearson correlation coeffitient for scoring function of representation based sequence search & alignment and it improved my score approximately 0.14 compared to blastn search (70% confidence. There were complicated issues).\n\nMy last submission (0.671) used dot product for scoring function + fragment masking (gappy short alignment regions were masked) and it improved my score around 0.07. I don't know which factor was critical.\n\nIn my previous experience, dot product was not so good thus I submitted it just for trial.\n\n\nUpdate (2025/06/17)\nI uploaded my presentation slide in my discussion page.\nPlease feel free to post comments if you have any questions.\nhttps://www.kaggle.com/competitions/stanford-rna-3d-folding/discussion/582295", "3245661": "Can you clarify? I wasn't able to form a submission by the first deadline but I have a solution that I am sure will make the leaderboard I just have to complete work on it. Am I still able to make a final submission and compete for the September deadline?", "3235093": "@rhijudas Are you going to publish interim results in near future?", "3218228": "Hi Rhiju,\nThank you for this incredible opportunity to continue testing and refining our solutions during this extended evaluation period!\nI'm really excited about the chance to further develop and validate the approach I've been working on. This additional time is invaluable for pushing the boundaries of what's possible in RNA structure prediction.", "3213817": "@rhijudas If we discover an interesting approach that performs better than the current public leaderboard, will it count toward the final winning list?", "3213796": "However, the competition information now shows: \"Submissions currently disabled,\" which means that from now on, it is no longer possible to make submissions. So regarding what you mentioned: \"If you get new ideas now after Final Submission Deadline, remember that you can still submit notebooks for scoring. Your scores won’t show up on leaderboards but if you find something interesting, please post in the Discussion.\"—how can this be done?", "3213821": "" } }