Processing of Achilles gene effect dependency matrix for computation of common essential genes with CoRe R package

Dear DepMap community,

I hope my message finds you well and healthy !! Based on a previous post regarding the exploitation of DepMap portal data, aiming to identify dependencies in specific cancer cell lines groups (Identify conditionally essential genes in clinical subgroups of a specific cancer type)

I was searching about various statistical methodologies to compute common essential genes for my cell lines/groups of interest, in order then as a second step to find the context-specific remaining essential genes, and to further validate initial approaches through using the portal;

On this premise, I found the relative R package CoRe (CoRe: a robustly benchmarked R package for identifying core-fitness genes in genome-wide pooled CRISPR-Cas9 screens | BMC Genomics | Full Text) which includes a specific function called FiPer, for computation of common essential genes; As I got the following feedback from the authors:

"The FiPer method works directly on a quantitative matrix of log fold changes/dependencies. The only preprocessing step I would recommend is to scale the essentiality profiles cell-wise using the CoRe.scale_to_essentials function. For this function to work, you need a character vector of a priori known essential (BAGEL_essential and BAGEL_nonEssential available as data object in the package) and nonessential genes and unscaled matrix. This is a normalization procedure to obtain the median of those essential genes at -1 and the median of those non-essential at 0"

Thus, my main question is the following: does the latest Achilles_gene_effect.csv matrix been scaled or processed in a similar way? From a small attached histogram, it seems that is has been similarly transformed, but I wanted to be certain that I could utilize the input as it is?


Thank you in advance for your time and consideration :slight_smile:

Kind Regards,


Hi Efstathios,

Chronos data has not been scaled in quite the same way, but the subtle difference shouldn’t impede that function. In fact, I believe the FiPeR method is an R implementation of the method we are already using to generate the file Achilles_common_essentials, so it should give a very similar list of genes (modulo the default parameters and some implementation details).



1 Like

Dear Joshua,

thank you very much for your response and feedback !!