{
  "id": 407048,
  "title": "About this competition",
  "url": "/competitions/cafa-5-protein-function-prediction/discussion/407048",
  "author_name": "",
  "post_date": "2023-05-05T00:35:57.803850Z",
  "votes": 3,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello. I have checked the evaluation section and am not sure why the submission file format is the way it is for this task. It seems to me that it is possible that a GO term that is considered relevant in one sub-ontology may not be relevant in another sub-ontology. So I thought it would be more natural to prepare three models for the three sub-ontologies and submit the predictions in separate submission files for each. Please forgive me if there are any misunderstandings.</p>",
  "messages": [
    {
      "id": "2246183",
      "postDate": "05/05/2023 00:35:57",
      "content": "<p>Hello. I have checked the evaluation section and am not sure why the submission file format is the way it is for this task. It seems to me that it is possible that a GO term that is considered relevant in one sub-ontology may not be relevant in another sub-ontology. So I thought it would be more natural to prepare three models for the three sub-ontologies and submit the predictions in separate submission files for each. Please forgive me if there are any misunderstandings.</p>",
      "rawMarkdown": "Hello. I have checked the evaluation section and am not sure why the submission file format is the way it is for this task. It seems to me that it is possible that a GO term that is considered relevant in one sub-ontology may not be relevant in another sub-ontology. So I thought it would be more natural to prepare three models for the three sub-ontologies and submit the predictions in separate submission files for each. Please forgive me if there are any misunderstandings.",
      "votes": null
    },
    {
      "id": "2246197",
      "postDate": "05/05/2023 00:51:56",
      "content": "<blockquote>\n  <p>It seems to me that it is possible that a GO term that is considered relevant in one sub-ontology may not be relevant in another sub-ontology.</p>\n</blockquote>\n<p>Hi. Each GO identifier (<code>GO:0000001</code>, <code>GO:0000002</code>, etc.) is associated with only one specific sub-ontology (BP, CC, or MF). Therefore, they can be submitted in a single file and then mapped to their respective sub-ontologies. Of course, nothing is preventing you from training three separate models and then merging their predictions into one submission file. Having a single submission file probably just makes it easier to evaluate with their script.</p>",
      "rawMarkdown": "> It seems to me that it is possible that a GO term that is considered relevant in one sub-ontology may not be relevant in another sub-ontology.\n\nHi. Each GO identifier (`GO:0000001`, `GO:0000002`, etc.) is associated with only one specific sub-ontology (BP, CC, or MF). Therefore, they can be submitted in a single file and then mapped to their respective sub-ontologies. Of course, nothing is preventing you from training three separate models and then merging their predictions into one submission file. Having a single submission file probably just makes it easier to evaluate with their script.",
      "votes": null
    },
    {
      "id": "2246238",
      "postDate": "05/05/2023 02:23:08",
      "content": "<p>Thank you for your answer. I will look into the submission file a little more.</p>",
      "rawMarkdown": "Thank you for your answer. I will look into the submission file a little more.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2246197,
      "author_name": "norixmb",
      "author_url": "",
      "post_date": "05/05/2023 00:51:56",
      "content": "<blockquote>\n  <p>It seems to me that it is possible that a GO term that is considered relevant in one sub-ontology may not be relevant in another sub-ontology.</p>\n</blockquote>\n<p>Hi. Each GO identifier (<code>GO:0000001</code>, <code>GO:0000002</code>, etc.) is associated with only one specific sub-ontology (BP, CC, or MF). Therefore, they can be submitted in a single file and then mapped to their respective sub-ontologies. Of course, nothing is preventing you from training three separate models and then merging their predictions into one submission file. Having a single submission file probably just makes it easier to evaluate with their script.</p>",
      "votes": null,
      "replies": [
        {
          "id": 2246238,
          "author_name": "ariafschneider",
          "author_url": "",
          "post_date": "05/05/2023 02:23:08",
          "content": "<p>Thank you for your answer. I will look into the submission file a little more.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2246183": "Hello. I have checked the evaluation section and am not sure why the submission file format is the way it is for this task. It seems to me that it is possible that a GO term that is considered relevant in one sub-ontology may not be relevant in another sub-ontology. So I thought it would be more natural to prepare three models for the three sub-ontologies and submit the predictions in separate submission files for each. Please forgive me if there are any misunderstandings.",
    "2246197": "> It seems to me that it is possible that a GO term that is considered relevant in one sub-ontology may not be relevant in another sub-ontology.\n\nHi. Each GO identifier (`GO:0000001`, `GO:0000002`, etc.) is associated with only one specific sub-ontology (BP, CC, or MF). Therefore, they can be submitted in a single file and then mapped to their respective sub-ontologies. Of course, nothing is preventing you from training three separate models and then merging their predictions into one submission file. Having a single submission file probably just makes it easier to evaluate with their script.",
    "2246238": "Thank you for your answer. I will look into the submission file a little more."
  },
  "source": "meta"
}