When I use "mageck count", I get a hesitation that lfc for positive selection pull from the negative selection cutoffs. This is most apparent when the output table is sorted by positive selection.
In 2017, the author answered a similar question that solved through updating to the latest version.But my software version is 0.5.9.2 which is much higher than the version at that time.
I observed the gene SLC22A17 's sgRNA have two FALSE and one TRUE in high_in_treatment, and the TRUE 's pvalue is the most significant. I guess it is because the TRUE sgRNA influenced the rank in the RRA algorithm. But what kind of selection should I define for gene SLC22A17. Is there any other better solution.Thanks for help.
Table 1. gene_summary. [The "pos|lfc" of gene which rank is 19/26/29 is negative ]
| id | num | pos|score | pos|p-value | pos|fdr | pos|rank | pos|goodsgrna | pos|lfc |
|---|---|---|---|---|---|---|---|
| SLC35A1 | 3 | 1.77E-09 | 2.40E-07 | 0.002475 | 1 | 3 | 9.743 |
| VMA21 | 3 | 2.38E-08 | 2.40E-07 | 0.002475 | 2 | 2 | 6.5683 |
| JAG1 | 3 | 1.71E-05 | 5.53E-05 | 0.315594 | 3 | 3 | 3.1071 |
| SLC22A17 | 3 | 0.00031586 | 0.00092729 | 0.88592 | 19 | 1 | -6.5016 |
| YPEL2 | 3 | 0.00046162 | 0.0013556 | 0.88592 | 26 | 1 | -2.0106 |
| IKZF1 | 3 | 0.00051021 | 0.0014768 | 0.88592 | 29 | 1 | -5.9793 |
Table 2. sgRNA_summary. [Gene SLC22A17 which rank is 19 ]
| sgrna | Gene | control_count | treatment_count | control_mean | treat_mean | LFC | control_var |
|---|---|---|---|---|---|---|---|
| HGLibA_44559 | SLC22A17 | 1.4111 | 0/1317 | 1.4111 | 658.5 | 8.0955 | 8.67E+05 |
| HGLibA_44557 | SLC22A17 | 90.313 | 0/0 | 90.313 | 0 | -6.5127 | 0 |
| HGLibA_44558 | SLC22A17 | 89.607 | 0/0 | 89.607 | 0 | -6.5016 | 0 |
| adj_var | score | p.low | p.high | p.twosided | FDR | high_in_treatment | |
| 116.72 | 60.82 | 1 | 0 | 0 | 0 | TRUE | |
| 20740 | 2.8924 | 0.0019116 | 0.99809 | 0.0038231 | 0.018508 | FALSE | |
| 20538 | 2.8902 | 0.0019248 | 0.99808 | 0.0038495 | 0.018508 | FALSE |
I noticed the parameter "--gene-lfc-method" of MaGeCK has the options "alphamedian/alphamean".When set this parameter, MaGeCK will only use the gene which fold change were negative to selected negative gene. Perhaps, this method of circumventing the opposite trend sgRNA may be made sense.