Entering edit mode
11 months ago
teresa
•
0
Hi,everyone. When I used MAGeCK count command to deal with my CRISPR screening results,I found that there are 1000 non-target-sgRNAs in my download Human_GeCKOv2_Library_combine.csv file,whether I should remove them or not before using the count command?Could anyone tell me?Because if I don't remove these non-target-sgRNAs,they will be in my final gene rank list after the test command. Thanks!
I would count them. You can then choose to filter (or not) them before the actual analysis. I find NTCs valuable to estimate an empirical false discovery rate, because after all these should not have any selection effect but (some will because of off-targets and other biases) so that can be used to get a "feel" for the potential side effects that might affect all other sgRNAs/genes.
Also, you can tell MAGeCK to use these NTCs as controls. If so then it will use them to make the Null distribution for its permutation-based testing. If no controls are used then all sgRNAs are used to build the Null. For large screens afaik using all genes is fine because typically most sgRNAs have no effect, hence the global distribution represents the "neutral" noise of the assay. For focused screens you should imho definitely have a good number of controls to test against.
Thank you so much！