# cellMCD Effectively Discovers Drug Resistance and Sensitivity Genes for Acute Myeloid Leukemia

## TL;DR

This paper introduces cellMCD, a new method that better identifies genes linked to drug resistance in acute myeloid leukemia by combining multiple omics data.

## Contribution

The paper introduces cellMCD, a novel gene prioritization method that outperforms existing approaches in integrating multi-omics data for drug response analysis.

## Key findings

- cellMCD performs as well or better than Fisher’s and SSz methods in simulations.
- cellMCD showed advantages in a CRISPR drug screen for acute myeloid leukemia.
- cellMCD is available as an R package on GitHub for broader use.

## Abstract

Background: Rapid advances in biotechnology provide researchers with the opportunity to integrate omics profiles (genomics, epigenomics, transcriptomics, proteomics, etc.) with multiple phenotypes or experimental conditions. In cancers such as acute myeloid leukemia (AML), where combination therapies are standard of care, identifying genetic drivers of drug resistance requires evaluating how genes are associated with multiple drug response phenotypes. Statistical analyses associating omics profiles with multiple phenotypes yield multiple significance values and rankings for each of many genes. There is a great need to consolidate these multiple rankings into a consensus ranking to prioritize specific genes for detailed follow-up wet-lab or clinical studies. Methods/Results: Here, we evaluate the well-known Fisher’s method, the sum of squared z-statistics (SSz), and the recently published cellMCD method as tools for gene prioritization. In simulation studies, cellMCD showed very similar or highly superior performance to the widely used Fisher’s and SSz methods. These advantages were also observed in an example application involving a CRISPR drug screen of an acute myeloid leukemia cell line. Conclusions: In summary, our results indicate that cellMCD should be more widely used for prioritizing discoveries from multiple omic association studies. These methods are available as an R package on github.