Hi, I ran both MAGECK RRA and MAGECK MLE on my CRISPRi screen data set where I have samples:
- DMSO control day 0
- DMSO control day 18 (end of experiment)
- drug 1 day 18
- drug 2 day 18
- drug 3 day 18
- drug 4 day 18
My goal is to find genes which knock-down sensitizes cells to drug treatment. However, many common essential genes which have negative beta scores, as expected, in the control condition have positive beta scores in all drug-treated samples. This means that the essential genes look like resistors to the drug which is most probably not a true biological effect. Is this caused by the MAGECK algorithm? Can this be prevented? I don't see this happening when I simply calculate log-fold changes from the guide RNA read-counts between conditions. Of note, I use IC50 drug doses which have an effect on cell growth, hence the number of cell doublings are different across conditions.
Can someone please help?
RRA command: mageck test -k /home/ubuntu/sgrna_count.txt -t drug1_Day18 -c DMSO_Day18 -n drug1 --control-sgrna /home/ubuntu/non_targeting_guides.txt --pdf-report
MLE command: mageck mle -k /home/ubuntu/sgrna_count.txt -d /home/ubuntu/design_matrix.txt -n /home/ubuntu/MAGECK_MLE/beta --control-sgrna /home/ubuntu/non_targeting_guides.txt
