The following is a summary of “Predicting pathways for old and new metabolites through clustering,” published in the December 2023 issue of Pediatrics by Siddharth et al.

The intricate web of metabolic pathways serves as a foundational aspect across all life forms, enabling energy extraction, synthesizing essential components, generating molecules crucial for environmental interactions, and detoxifying processes. However, while the ongoing discovery of novel metabolites and pathways persists, predicting pathways for these new metabolites remains challenging. 

The intricate nature of elucidating pathways for new metabolites often demands an extensive investment of time, leading to a situation where only approximately 60% of known metabolites are attributed to specific pathways, per the Human Metabolome Database (HMDB) records. This study introduces an innovative approach centered on metabolite structure for pathway identification. Utilizing SMILES annotations, researchers extracted 201 distinct features and cross-referenced them with data from PubMed abstracts and HMDB to pinpoint new metabolites. 

Employing clustering algorithms on both groups of features, the study group scrutinized the correlations among metabolites and observed that these clusters successfully linked approximately 92% of recognized metabolites to their corresponding pathways. Consequently, this methodology showcases promising potential in accurately predicting metabolic pathways for newly discovered metabolites, offering a valuable tool in the ongoing pursuit of understanding complex metabolic networks.

Source: sciencedirect.com/science/article/abs/pii/S0022519323002813