Applying machine learning to identify ionizable lipids for nanoparticle-mediated delivery of mRNA

mRNA lipid nanoparticles (LNPs) have a tremendous potential to treat, cure, or prevent many diseases. To identify promising candidates for each application, most studies screen dozens to hundreds of formulations with ionizable lipids synthesized using a single type of chemistry. However, this technique leaves the ionizable lipids synthesized through multi-step chemistries underexplored. This gap in the repertoire of structures is of particular significance because it affects the screening of analogs that are structurally similar to SM-102 and ALC-0315, the ionizable lipids used to formulate the clinically approved mRNA LNP COVID-19 vaccines. Herein, we address this by employing LightGBM, a machine learning algorithm, to reduce the burden of screening these types of ionizable lipids by learning from the breadth and diversity of lipids that have already been tested. We first evaluate the ability of LightGBM to predict LNP potency across heterogeneous chemistries from different studies to achieve an R2 of 0.94. After establishing the predictive capacity of the model, we then identify the number of outside carbons in the ionizable lipid as the most important factor contributing to transfection efficiency. From this finding, we subsequently apply the algorithm to predict the effect of formulating nanoluciferase mRNA LNPs using analogs of SM-102 and ALC-0315 with small changes in the number of outside carbons on luciferase activity in HEK293T cells and achieve an R2 of 0.83. Importantly, this correlation encompasses novel lipids not included within the database used to train the algorithm. Overall, this study demonstrates the potential of machine learning to accelerate the development of new ionizable lipids by simplifying the screening process. ### Competing Interest Statement The authors have declared no competing interest.