{
  "id": 453475,
  "title": "Predicting Bulk Data + Limma Pipeline",
  "url": "/competitions/open-problems-single-cell-perturbations/discussion/453475",
  "author_name": "LBSDXQ",
  "post_date": "2023-11-06T13:42:52.652000",
  "votes": 3,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I would like to know which prediction pipeline you guys chose. In my practice, directly predicting the DE values according to cell types and SM names gives better results. However, I personally feel that first predicting the bulk count data followed by Limma would be a more robust and practical solution. Unfortunately, I got inferior results following the second pipeline. Could anyone share your results and opinions about the two different pipelines?</p>",
  "messages": [
    {
      "id": 2514739,
      "postDate": "2023-11-06T13:42:52.653Z",
      "content": "<p>I would like to know which prediction pipeline you guys chose. In my practice, directly predicting the DE values according to cell types and SM names gives better results. However, I personally feel that first predicting the bulk count data followed by Limma would be a more robust and practical solution. Unfortunately, I got inferior results following the second pipeline. Could anyone share your results and opinions about the two different pipelines?</p>",
      "rawMarkdown": "I would like to know which prediction pipeline you guys chose. In my practice, directly predicting the DE values according to cell types and SM names gives better results. However, I personally feel that first predicting the bulk count data followed by Limma would be a more robust and practical solution. Unfortunately, I got inferior results following the second pipeline. Could anyone share your results and opinions about the two different pipelines?",
      "votes": 3
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2514739": "I would like to know which prediction pipeline you guys chose. In my practice, directly predicting the DE values according to cell types and SM names gives better results. However, I personally feel that first predicting the bulk count data followed by Limma would be a more robust and practical solution. Unfortunately, I got inferior results following the second pipeline. Could anyone share your results and opinions about the two different pipelines?"
  }
}