{
  "id": 406504,
  "title": "Validation Strategy",
  "url": "/competitions/cafa-5-protein-function-prediction/discussion/406504",
  "author_name": "",
  "post_date": "2023-05-02T15:45:40.170379200Z",
  "votes": 10,
  "comment_count": 1,
  "views": 0,
  "content": "<p>What has been a good validation strategy in the past CAFA competitions? I'm looking at this repo <a href=\"https://github.com/Rostlab/goPredSim/tree/master/data/goa_annotations\" target=\"_blank\">https://github.com/Rostlab/goPredSim/tree/master/data/goa_annotations</a> highlighted a while ago by <a href=\"https://www.kaggle.com/Tilii\" target=\"_blank\">@Tilii</a>. It provides experimental goa for 2022. I was thinking of using the experimental GO annotations from the past (2022) and evaluating on new annotations present in the competition train set, to get an idea of our model performance.</p>",
  "messages": [
    {
      "id": "2242960",
      "postDate": "05/02/2023 15:45:40",
      "content": "<p>What has been a good validation strategy in the past CAFA competitions? I'm looking at this repo <a href=\"https://github.com/Rostlab/goPredSim/tree/master/data/goa_annotations\" target=\"_blank\">https://github.com/Rostlab/goPredSim/tree/master/data/goa_annotations</a> highlighted a while ago by <a href=\"https://www.kaggle.com/Tilii\" target=\"_blank\">@Tilii</a>. It provides experimental goa for 2022. I was thinking of using the experimental GO annotations from the past (2022) and evaluating on new annotations present in the competition train set, to get an idea of our model performance.</p>",
      "rawMarkdown": "What has been a good validation strategy in the past CAFA competitions? I'm looking at this repo https://github.com/Rostlab/goPredSim/tree/master/data/goa_annotations highlighted a while ago by @Tilii. It provides experimental goa for 2022. I was thinking of using the experimental GO annotations from the past (2022) and evaluating on new annotations present in the competition train set, to get an idea of our model performance.",
      "votes": null
    },
    {
      "id": "2243022",
      "postDate": "05/02/2023 16:33:54",
      "content": "<p>Previous succesful works mainly focused on predicting added annotations based on homology</p>",
      "rawMarkdown": "Previous succesful works mainly focused on predicting added annotations based on homology",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2243022,
      "author_name": "danofer",
      "author_url": "",
      "post_date": "05/02/2023 16:33:54",
      "content": "<p>Previous succesful works mainly focused on predicting added annotations based on homology</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2242960": "What has been a good validation strategy in the past CAFA competitions? I'm looking at this repo https://github.com/Rostlab/goPredSim/tree/master/data/goa_annotations highlighted a while ago by @Tilii. It provides experimental goa for 2022. I was thinking of using the experimental GO annotations from the past (2022) and evaluating on new annotations present in the competition train set, to get an idea of our model performance.",
    "2243022": "Previous succesful works mainly focused on predicting added annotations based on homology"
  },
  "source": "meta"
}