Diversity of FAAL enzymes and prediction of their substrate specificity using FAALPred

FAALs (Fatty Acyl-AMP Ligases) recruit and incorporate fatty acids during the biosynthesis of secondary metabolites. Their diversity, distribution and substrate specificity remain poorly understood, which limits functional predictions from sequence data. In this study, we explored the prevalence and diversity of FAAL enzymes across the tree of life and show that FAALs are widely distributed among bacteria, with distinct clades associated with specific taxonomic groups and/or biosynthetic pathways. Specifically, bacterial FAALs were found to be predominantly associated with type I PKS, NRPS or NPRS-PKS hybrid biosynthetic pathways. The phylogenetic placement of FAALs was not correlated to the chain length of the fatty acids that they activate and load. Therefore, we developed a deep learning-based prediction algorithm (FAALPred) to forecast the chain length of the fatty acid substrate of a given FAAL sequence. The robustness and accuracy of the predictions generated by FAALPred were validated using independent in vitro and in silico data. We anticipate that FAALPred will not only accelerate secondary metabolite structural predictions and subsequent discovery from FAAL-associated pathways, but also facilitate metabolic engineering of lipoylation.