{
  "id": 356222,
  "title": "Practice in single cell data analysis",
  "url": "/competitions/open-problems-multimodal/discussion/356222",
  "author_name": "",
  "post_date": "2022-09-29T18:48:19.181275300Z",
  "votes": 6,
  "comment_count": 1,
  "views": 0,
  "content": "<p>When bioinformaticians do ananysis on single-cell data the preprocessing step usually follow:<br>\na. Filtering out low quality cells, i.e. dead cells and doublet-cells (two cells are loaded at the same time, so not single-cell resolution)<br>\nb. Normalise the data: normalized against library size (already done by competition host)<br>\nc. Find variable features (dimension reduction/feature selection): find genes with varaible expression between cells<br>\nd. Scale the features</p>\n<p>I wonder whether these are helpful in machine learning.</p>\n<p>And do the host perform cell filtering? It seems to me there is no obvious outlier cells when I do EDA.</p>",
  "messages": [
    {
      "id": "1962537",
      "postDate": "09/29/2022 18:48:19",
      "content": "<p>When bioinformaticians do ananysis on single-cell data the preprocessing step usually follow:<br>\na. Filtering out low quality cells, i.e. dead cells and doublet-cells (two cells are loaded at the same time, so not single-cell resolution)<br>\nb. Normalise the data: normalized against library size (already done by competition host)<br>\nc. Find variable features (dimension reduction/feature selection): find genes with varaible expression between cells<br>\nd. Scale the features</p>\n<p>I wonder whether these are helpful in machine learning.</p>\n<p>And do the host perform cell filtering? It seems to me there is no obvious outlier cells when I do EDA.</p>",
      "rawMarkdown": "When bioinformaticians do ananysis on single-cell data the preprocessing step usually follow:\na. Filtering out low quality cells, i.e. dead cells and doublet-cells (two cells are loaded at the same time, so not single-cell resolution)\nb. Normalise the data: normalized against library size (already done by competition host)\nc. Find variable features (dimension reduction/feature selection): find genes with varaible expression between cells\nd. Scale the features\n\nI wonder whether these are helpful in machine learning.\n\nAnd do the host perform cell filtering? It seems to me there is no obvious outlier cells when I do EDA.",
      "votes": null
    },
    {
      "id": "1962816",
      "postDate": "09/30/2022 02:16:39",
      "content": "<p>Thank you <a href=\"https://www.kaggle.com/jinyang18\" target=\"_blank\">@jinyang18</a> </p>",
      "rawMarkdown": "Thank you @jinyang18",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1962816,
      "author_name": "saberghaderi",
      "author_url": "",
      "post_date": "09/30/2022 02:16:39",
      "content": "<p>Thank you <a href=\"https://www.kaggle.com/jinyang18\" target=\"_blank\">@jinyang18</a> </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1962537": "When bioinformaticians do ananysis on single-cell data the preprocessing step usually follow:\na. Filtering out low quality cells, i.e. dead cells and doublet-cells (two cells are loaded at the same time, so not single-cell resolution)\nb. Normalise the data: normalized against library size (already done by competition host)\nc. Find variable features (dimension reduction/feature selection): find genes with varaible expression between cells\nd. Scale the features\n\nI wonder whether these are helpful in machine learning.\n\nAnd do the host perform cell filtering? It seems to me there is no obvious outlier cells when I do EDA.",
    "1962816": "Thank you @jinyang18"
  },
  "source": "meta"
}