This study is centered on Natural Products (NPs) - specific chemicals synthesized by living organisms. These NPs hold significant importance in various domains, notably medicine, agriculture, and ecology. A primary resource for our research is the LOTUS database, which catalogues a vast array of NPs and their occurrence. Yet, a gap exists: there are no existing model to predict the occurrence of these NPs across different species.

In our initial strategy, the occurrence of natural products was viewed as a collection of observations and their associated variables. Although simple, this strategy immediately showed its limits when dealing with the complex nature of NPs. We switched to an advanced graph-based method after seeing the necessity for a more thorough strategy to accurately represent the intricate interactions governing NPs expression. When considering species in a phylogeny or molecular pathways, the graph-based method perceives data as a network of connected entities, offering a far more logical and natural way of thinking. By employing this better methodology, we have developed a more effective approach for investigating the intricate world of Natural Products. We hope that this strategy will open up new research directions and possibly result in ground-breaking NP-related findings.