Whilst my bio knowledge is slim, one thing that I'm aware of is that even in the best commercial single cell RNA sequencers, you only get up to 30% coverage. Meaning, of the RNA floating around in a cell, you'll only catch 30% of them in a single sequencing attempt. You can of course batch a bunch of cells together, but then you no longer have a snapshot, rather an average of expression.
\nSimilar issue with proteins, I know you can sequence peptides using mass spectrometry, but that has its own issues including different results with different instruments.
\nBasically, I'm wondering what the limitations of the DNA/RNA/Protein sequencing in this dataset are? I'm thinking this may be helpful when designing models?
", "messages": [ { "id": "1915019", "postDate": "08/26/2022 15:59:20", "content": "Whilst my bio knowledge is slim, one thing that I'm aware of is that even in the best commercial single cell RNA sequencers, you only get up to 30% coverage. Meaning, of the RNA floating around in a cell, you'll only catch 30% of them in a single sequencing attempt. You can of course batch a bunch of cells together, but then you no longer have a snapshot, rather an average of expression.
\nSimilar issue with proteins, I know you can sequence peptides using mass spectrometry, but that has its own issues including different results with different instruments.
\nBasically, I'm wondering what the limitations of the DNA/RNA/Protein sequencing in this dataset are? I'm thinking this may be helpful when designing models?
", "rawMarkdown": "Whilst my bio knowledge is slim, one thing that I'm aware of is that even in the best commercial single cell RNA sequencers, you only get up to 30% coverage. Meaning, of the RNA floating around in a cell, you'll only catch 30% of them in a single sequencing attempt. You can of course batch a bunch of cells together, but then you no longer have a snapshot, rather an average of expression.\n\nSimilar issue with proteins, I know you can sequence peptides using mass spectrometry, but that has its own issues including different results with different instruments.\n\n**Basically, I'm wondering what the limitations of the DNA/RNA/Protein sequencing in this dataset are? I'm thinking this may be helpful when designing models?**", "votes": null }, { "id": "1915095", "postDate": "08/26/2022 16:51:34", "content": "These are like real life dropout in information we sample….so the real life data is noisy in principle :)
", "rawMarkdown": "These are like real life dropout in information we sample....so the real life data is noisy in principle :)", "votes": null }, { "id": "1915116", "postDate": "08/26/2022 17:14:57", "content": "On the protein side, my understanding of Cite-seq is that the proteins are NOT sequenced using mass-spec, rather there is a panel of barcoded antibodies that stick to surface proteins on the cell. This means that you can only really sense 1) surface proteins 2) proteins that already have complementary antibodies in the Cite-seq panel.
\nThus I could see there being some disconnect where RNA is detected, but it might have nothing to do with the proteins that have been detected since protein detection is much more limited.
\nI'm not an expert though, this is all just from the wiki which I found very helpful, especially the limitations section at the end: https://en.wikipedia.org/wiki/CITE-Seq
", "rawMarkdown": "On the protein side, my understanding of Cite-seq is that the proteins are NOT sequenced using mass-spec, rather there is a panel of barcoded antibodies that stick to surface proteins on the cell. This means that you can only really sense 1) surface proteins 2) proteins that already have complementary antibodies in the Cite-seq panel. \n\nThus I could see there being some disconnect where RNA is detected, but it might have nothing to do with the proteins that have been detected since protein detection is much more limited.\n\nI'm not an expert though, this is all just from the wiki which I found very helpful, especially the limitations section at the end: https://en.wikipedia.org/wiki/CITE-Seq", "votes": null }, { "id": "1916038", "postDate": "08/27/2022 15:38:00", "content": "The optimist's way of looking at it haha 😅
", "rawMarkdown": "The optimist's way of looking at it haha 😅", "votes": null } ], "comments": [ { "id": 1915095, "author_name": "drpatrickchan", "author_url": "", "post_date": "08/26/2022 16:51:34", "content": "These are like real life dropout in information we sample….so the real life data is noisy in principle :)
", "votes": null, "replies": [ { "id": 1916038, "author_name": "subachev", "author_url": "", "post_date": "08/27/2022 15:38:00", "content": "The optimist's way of looking at it haha 😅
", "votes": null, "replies": [] } ] }, { "id": 1915116, "author_name": "lukeschiefelbein", "author_url": "", "post_date": "08/26/2022 17:14:57", "content": "On the protein side, my understanding of Cite-seq is that the proteins are NOT sequenced using mass-spec, rather there is a panel of barcoded antibodies that stick to surface proteins on the cell. This means that you can only really sense 1) surface proteins 2) proteins that already have complementary antibodies in the Cite-seq panel.
\nThus I could see there being some disconnect where RNA is detected, but it might have nothing to do with the proteins that have been detected since protein detection is much more limited.
\nI'm not an expert though, this is all just from the wiki which I found very helpful, especially the limitations section at the end: https://en.wikipedia.org/wiki/CITE-Seq
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1915019": "Whilst my bio knowledge is slim, one thing that I'm aware of is that even in the best commercial single cell RNA sequencers, you only get up to 30% coverage. Meaning, of the RNA floating around in a cell, you'll only catch 30% of them in a single sequencing attempt. You can of course batch a bunch of cells together, but then you no longer have a snapshot, rather an average of expression.\n\nSimilar issue with proteins, I know you can sequence peptides using mass spectrometry, but that has its own issues including different results with different instruments.\n\n**Basically, I'm wondering what the limitations of the DNA/RNA/Protein sequencing in this dataset are? I'm thinking this may be helpful when designing models?**", "1915095": "These are like real life dropout in information we sample....so the real life data is noisy in principle :)", "1915116": "On the protein side, my understanding of Cite-seq is that the proteins are NOT sequenced using mass-spec, rather there is a panel of barcoded antibodies that stick to surface proteins on the cell. This means that you can only really sense 1) surface proteins 2) proteins that already have complementary antibodies in the Cite-seq panel. \n\nThus I could see there being some disconnect where RNA is detected, but it might have nothing to do with the proteins that have been detected since protein detection is much more limited.\n\nI'm not an expert though, this is all just from the wiki which I found very helpful, especially the limitations section at the end: https://en.wikipedia.org/wiki/CITE-Seq", "1916038": "The optimist's way of looking at it haha 😅" }, "source": "meta" }