Do we need to filter out the 0 counts sgRNA in CRISPR analysis?

Question

[leranwangcs]

Hi,

I'm doing CRISPR analysis and there is a question, when we perform CRISPR analysis, is there a step to remove low count sgRNAs from control and/or treatment samples?

The reason of this question is that, if sgRNA(A) has a very low control count (perhaps 0) and a moderate treatment count (perhaps 500), it will have a higher logFC than sgRNA(B) which has a moderate control count (perhaps 500) and a high treatment count (perhaps 10000). However, biologically speaking, sgRNA(B) should be more interesting, as the treatment count is very high and it takes up a bigger portion of the treatment (sorted) sample sgRNA pool.

Is there any suggestion on this problem?

Thanks! Leran

Answer 1

[xiaoguang]

Are you doing crispr_screening? I know that the software mageck will remove 0 read sgRNA firstly

[leranwangcs]

Hi Xiaoguang,

Thanks for asking. I tried several tools including magick, I know there is a perimeter in magick that allows you to remove 0s, but I'm not sure if this is a must step in crispr analysis.

Leran

[xiaoguang]

You are right. Filtering is a necessary step before any data processing.