I am interested in a class of methyltransferase genes, and, based on my experience when looking up these, I have a question regarding scores versus classification (as essential or selective). I downloaded the DepMap data for these 117 genes, and I then took the median of the score ( (DepMap Public 23Q2+Score, Chronos) for each gene over all the 1095 cell lines, and ranked the genes according to these median values. I next looked at the top (highest negative score) genes, regarding how they had been classified, i.e. as “Commonly essential” (CE), “Strongly selective” (SS), or no label. I then, as expected, saw that all the top genes were characterized as CE. However, there was also one case where a gene (METTL14) had a very high score, much higher than other genes that were classified as CE or SS, but still got none of these “labels” (see attached picture). I understand that some “discrepancies” should be expected when comparing these parameters, but I still found the “METTL14 case” quite extreme. I therefore have two questions: 1) Is METTL14 correctly annotated? 2) I wish to make an overview figure (for a review article) showing to what extent these genes are required for cell growth, and wonder if you have recommendations on which parameters to use/indicate (score and/or annotation; I planned to indicate both)?
Hello pfalnes,
We define common essential status by gene effect rank in the 90th percentile least dependent line. The emphasis is on the consistency of the dependency rather than strength. It’s theoretically possible for a gene to have a very negative mean but also a large population of non-dependent cell lines. METTL14 is on the edge, and in fact in the internal version of the data it does tip over into common essential. But I wouldn’t describe it as extreme; it’s a pretty weak dependency overall.
What you show in the paper really comes down to what you want to emphasize. I can imagine different versions depending on whether you want to show that they are consistently required for viability, or instead focus on a population that more strongly responds to KO.
Hello Joshua,
Thanks for explaining this so well. To me it appeared somewhat counterintuitive that METTL14 was indicated as being essential in about 70 % (767/1095) of the cell lines but still was not catagorized as being “Commonly essential”. But I now of course realize that this is due to how the criteria are defined. I suppose one may imagine that a subpopulation of the cell lines somehow has acquired independence of METTL14 (and that there could be some interesting biology here…).
Best,
Pål
Hello Joshua,
A small add-on to my previous email: To look for compensatory mechanisms, it may be of interest to look into which genes show an altered expression in the cell lines that are not dependent of METTL14, i.e. to identify genes whose expression correlate (or inversely correlate) with the CHRONOS score for METTL14 dependence. I guess this may be done by downloading and analysing the actual data (but my bioinformatics skills are crude). Can this be addressed through the tools at your portal? (maybe not, I could not find an obvious way to do this…).
Best,
Pål
