Why use MOAL
Resistance is a major barrier to cancer remission. However, it is not practical to test all the millions of potential drug combinations due to cost limitations. To help prioritize drug pairings for further evaluation, we studied the similarity of drugs grouped by mechanisms of action (MOA) in published drug screens. With this process, termed the Mechanism of Action Landscape (MOAL), 131 drug MOA pairings were prioritized from more than 1.8 million potential drug pairings. Though this method does not evaluate synergy, it has been used to study more than a thousand drugs whereas other methods have only studied on the order of ten to one hundred drugs. In addition to recovering established drug combinations (e.g., MEK and RAF inhibitors in skin cancer), this method has also detected potential drug pairings which have not been well studied and demonstrated that results can vary depending on tissue type. Furthermore, we show that the MOAL method is beneficial over a simple UMAP visualization because it uses statistics to eliminate more than 96% of potential drug MOA pairings from consideration and identifies significant MOA pairings which may not be obvious from their distance on a UMAP visualization (e.g., BET and HDAC inhibitors). With the MOAL R package, this method can also be used to evaluate the global performance of drug sensitivity prediction algorithms.
To learn more about DMEA, our previous algorithm which is the building block of MOAL, you can read our bioRxiv pre-print here: https://www.biorxiv.org/content/10.1101/2022.03.15.484520v3
To learn more about enrichment analysis in general, you can refer to this paper by Subramanian et al and watch this YouTube video: