While computing differential expression it makes sense to have some cells that are the control samples, against which the effect of any compound/drug is measured. But what does it mean to have a feature named \"control\" in de_train.parquet?
", "messages": [ { "id": "2458964", "postDate": "09/27/2023 23:57:35", "content": "While computing differential expression it makes sense to have some cells that are the control samples, against which the effect of any compound/drug is measured. But what does it mean to have a feature named \"control\" in de_train.parquet?
", "rawMarkdown": "While computing differential expression it makes sense to have some cells that are the control samples, against which the effect of any compound/drug is measured. But what does it mean to have a feature named \"control\" in de_train.parquet?", "votes": null }, { "id": "2458999", "postDate": "09/28/2023 01:03:43", "content": "The rows where control is True means that the chemical measured in that row is a control chemical.
\nThe negative control dimethyl sulfoxide (DMSO) is a common solvent and (in my understanding) is used as the solvent for all plates.
\nThe positive controls (dabrfenib and belinostat) are used as baselines.
The control in the de_train.parquet lablels compouns used as postive controls. My understanding is that, since they are known to cause perturbations, they are used for quality control to detect and maybe discard \"falase negative\" results from the same plate.
Think of those covid/preganancy test, you need the control bar to show up for the test to be valid.
\nAs for the relevance as training data, you can choose to include those data, or not to include them. Personally, I don't know which option is better.
", "rawMarkdown": "The `control` in the de_train.parquet lablels compouns used as postive controls. My understanding is that, since they are known to cause perturbations, they are used for quality control to detect and maybe discard \"falase negative\" results from the same plate. \n\nThink of those covid/preganancy test, you need the control bar to show up for the test to be valid. \n\nAs for the relevance as training data, you can choose to include those data, or not to include them. Personally, I don't know which option is better.", "votes": null }, { "id": "2459297", "postDate": "09/28/2023 06:24:42", "content": "Thank you both.
", "rawMarkdown": "Thank you both.", "votes": null }, { "id": "2459872", "postDate": "09/28/2023 12:36:52", "content": "This is correct. When you order compounds from a vendor, they arrive as a powder. To treat cells with a compound, you must first dissolve this powder in a solvent (also called a vehicle). Many compounds are not water-soluble, so they get dissolved in DMSO. This is the case for all compounds in this dataset.
\nWe include a negative control set that includes just the solvent (DMSO) with no compound added. This is also called a vehicle control.
", "rawMarkdown": "This is correct. When you order compounds from a vendor, they arrive as a powder. To treat cells with a compound, you must first dissolve this powder in a solvent (also called a vehicle). Many compounds are not water-soluble, so they get dissolved in DMSO. This is the case for all compounds in this dataset. \n\nWe include a negative control set that includes just the solvent (DMSO) with no compound added. This is also called a vehicle control.", "votes": null }, { "id": "2460566", "postDate": "09/28/2023 22:12:25", "content": "In this case, what exactly is the difference between positive and negative controls? Because all three DMSO, Dabrfenib and Belinostat seem to have been used as solvents. The first one for compounds and the other two for cells to establish baseline.
", "rawMarkdown": "In this case, what exactly is the difference between positive and negative controls? Because all three DMSO, Dabrfenib and Belinostat seem to have been used as solvents. The first one for compounds and the other two for cells to establish baseline.", "votes": null }, { "id": "2460581", "postDate": "09/28/2023 22:50:40", "content": "As Daniel wrote, DMSO is the solvent and is present in every well.
\nSay we are conducting an experiment on DrugX. We dissolve DrugX in DMSO and introduce it into the well.
\nIt's called the negative control, because it is the baseline for all effects. In other words, effect we measure caused by DrugX is really DrugX + DMSO.
\nDabrfenib and Belinostat are FDA approved cancer fighting medications, so extensive testing have already been conducted using them, and they are known to be effective.
\nIn other words, if in this experiment Dabrfenib or Belinostat do not produce their expected results, something is wrong with the experimental setup. That's why they are called positive controls.
I noticed that removing controls from the training dataset improves the quality of my models.
", "rawMarkdown": "I noticed that removing controls from the training dataset improves the quality of my models.", "votes": null } ], "comments": [ { "id": 2458999, "author_name": "laurasisson", "author_url": "", "post_date": "09/28/2023 01:03:43", "content": "The rows where control is True means that the chemical measured in that row is a control chemical.
\nThe negative control dimethyl sulfoxide (DMSO) is a common solvent and (in my understanding) is used as the solvent for all plates.
\nThe positive controls (dabrfenib and belinostat) are used as baselines.
This is correct. When you order compounds from a vendor, they arrive as a powder. To treat cells with a compound, you must first dissolve this powder in a solvent (also called a vehicle). Many compounds are not water-soluble, so they get dissolved in DMSO. This is the case for all compounds in this dataset.
\nWe include a negative control set that includes just the solvent (DMSO) with no compound added. This is also called a vehicle control.
", "votes": null, "replies": [ { "id": 2460566, "author_name": "arpitamhow", "author_url": "", "post_date": "09/28/2023 22:12:25", "content": "In this case, what exactly is the difference between positive and negative controls? Because all three DMSO, Dabrfenib and Belinostat seem to have been used as solvents. The first one for compounds and the other two for cells to establish baseline.
", "votes": null, "replies": [ { "id": 2460581, "author_name": "laurasisson", "author_url": "", "post_date": "09/28/2023 22:50:40", "content": "As Daniel wrote, DMSO is the solvent and is present in every well.
\nSay we are conducting an experiment on DrugX. We dissolve DrugX in DMSO and introduce it into the well.
\nIt's called the negative control, because it is the baseline for all effects. In other words, effect we measure caused by DrugX is really DrugX + DMSO.
\nDabrfenib and Belinostat are FDA approved cancer fighting medications, so extensive testing have already been conducted using them, and they are known to be effective.
\nIn other words, if in this experiment Dabrfenib or Belinostat do not produce their expected results, something is wrong with the experimental setup. That's why they are called positive controls.
The control in the de_train.parquet lablels compouns used as postive controls. My understanding is that, since they are known to cause perturbations, they are used for quality control to detect and maybe discard \"falase negative\" results from the same plate.
Think of those covid/preganancy test, you need the control bar to show up for the test to be valid.
\nAs for the relevance as training data, you can choose to include those data, or not to include them. Personally, I don't know which option is better.
", "votes": null, "replies": [] }, { "id": 2459297, "author_name": "arpitamhow", "author_url": "", "post_date": "09/28/2023 06:24:42", "content": "Thank you both.
", "votes": null, "replies": [] }, { "id": 2461281, "author_name": "mishazybin", "author_url": "", "post_date": "09/29/2023 12:01:24", "content": "I noticed that removing controls from the training dataset improves the quality of my models.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2458964": "While computing differential expression it makes sense to have some cells that are the control samples, against which the effect of any compound/drug is measured. But what does it mean to have a feature named \"control\" in de_train.parquet?", "2458999": "The rows where control is True means that the chemical measured in that row is a control chemical.\nThe negative control dimethyl sulfoxide (DMSO) is a common solvent and (in my understanding) is used as the solvent for all plates.\nThe positive controls (dabrfenib and belinostat) are used as baselines.", "2459016": "The `control` in the de_train.parquet lablels compouns used as postive controls. My understanding is that, since they are known to cause perturbations, they are used for quality control to detect and maybe discard \"falase negative\" results from the same plate. \n\nThink of those covid/preganancy test, you need the control bar to show up for the test to be valid. \n\nAs for the relevance as training data, you can choose to include those data, or not to include them. Personally, I don't know which option is better.", "2459297": "Thank you both.", "2459872": "This is correct. When you order compounds from a vendor, they arrive as a powder. To treat cells with a compound, you must first dissolve this powder in a solvent (also called a vehicle). Many compounds are not water-soluble, so they get dissolved in DMSO. This is the case for all compounds in this dataset. \n\nWe include a negative control set that includes just the solvent (DMSO) with no compound added. This is also called a vehicle control.", "2460566": "In this case, what exactly is the difference between positive and negative controls? Because all three DMSO, Dabrfenib and Belinostat seem to have been used as solvents. The first one for compounds and the other two for cells to establish baseline.", "2460581": "As Daniel wrote, DMSO is the solvent and is present in every well.\nSay we are conducting an experiment on DrugX. We dissolve DrugX in DMSO and introduce it into the well.\nIt's called the negative control, because it is the baseline for all effects. In other words, effect we measure caused by DrugX is really DrugX + DMSO.\nDabrfenib and Belinostat are FDA approved cancer fighting medications, so extensive testing have already been conducted using them, and they are known to be effective.\nIn other words, if in this experiment Dabrfenib or Belinostat do not produce their expected results, something is wrong with the experimental setup. That's why they are called positive controls.", "2461281": "I noticed that removing controls from the training dataset improves the quality of my models." }, "source": "meta" }