We compared the performance of four publicly available in silico fragmentation tools that help in structure elucidation of metabolites (MetFrag, CFM-ID, MAGMa+ and MS-FINDER).
Using the output of these tools we developed a voting/consensus model which is combining, re-ordering and re-ranking the candidates file and increases the correct hit percentage.
We used 520 compounds from the 2016 CASMI challenge with 312 compounds for training and 208 compounds for validation purposes.