Hi
@Laurence Kuhlburger, I'm not sure TBH how Mageck works or if it would be compatible with CRISPRi screens. I have not tested it myself or compared to our results. But I strongly guess it would also work and produce similar results. At least in the article you linked they used it with CRISPRi data, so why not. In the end, the task is quite simple, to find genes that reduce fitness in a particular environment compared to a set of controls. I guess different tools can come to roughly the same conclusions. Maybe some background to the history of our apporach: We developed the 'fitness score' already back in 2018 when we working a lot with a new CRISPRi library. I see that Mageck algorithm predates this, but sadly I wasn't aware of that. Anyway we published our results in Nat Comms in 2020, and have one more upcoming publication for which I can't change the fitness calculation anymore.
I have also compared our fitness score (DESeq2 based) to a scoring algorithm for barcoded Tn-Seq libraries (this one:
https://bitbucket.org/berkeleylab/feba/src/master/) that roughly produce the same type of data. The results were virtually identical. This makes me think that Mageck could absolutely be used for CRISPRi data. I could use it as a replacement for our fitness score in the future, or a complementation. You seem to be going ahead with mageck for this pipeline here, so I suggest once you have a pilot I can test it with some of our data and compare.
Apologies for lengthy post.