While catalytic reactions of biomolecular processes play an indispensable role in life, extracting the underlying molecular picture often remains challenging. Based on ab initio simulations of the self-cleavage reaction of hairpin ribozyme, mode-decomposed infrared spectra, and cosine similarity analysis to correlate the product with reactant IR spectra, we demonstrate a strategy to extract molecular details from characteristic spectral changes. Our results are in almost quantitative agreement with the experimental IR band library of nucleic acids and suggest that the spectral range of 800-1200 cm-1 is particularly valuable to monitor self-cleavage. Importantly, the cosine similarities also disclose that IR peaks subject to slight shifts due to self-cleavage might be unrelated, while strongly shifting resonances can correspond to the same structural dynamics. This framework of correlating complex IR spectra at the molecular level along biocatalytic reaction pathways is broadly applicable.