CFM-ID 3.0: Significantly Improved ESI-MS/MS Prediction and Compound Identification

Abstract

Metabolite identification for untargeted metabolomics is often hampered by the lack of experimentally collected reference spectra from tandem mass spectrometry (MS/MS). To circumvent this problem, Competitive Fragmentation Modeling-ID (CFM-ID) was developed to accurately predict electrospray ionization-MS/MS (ESI-MS/MS) spectra from chemical structures and to aid in compound identification via MS/MS spectral matching. While earlier versions of CFM-ID performed very well, CFM-ID’s performance for predicting the MS/MS spectra of certain classes of compounds, including many lipids, was quite poor. Furthermore, CFM-ID’s compound identification capabilities were limited because it did not use experimentally available MS/MS spectra nor did it exploit metadata in its spectral matching algorithm. Here, we describe significant improvements to CFM-ID’s performance and speed. These include (1) the implementation of a rule-based fragmentation approach for lipid MS/MS spectral prediction, which greatly improves the speed and accuracy of CFM-ID; (2) the inclusion of experimental MS/MS spectra and other metadata to enhance CFM-ID’s compound identification abilities; (3) the development of new scoring functions that improves CFM-ID’s accuracy by 21.1%; and (4) the implementation of a chemical classification algorithm that correctly classifies unknown chemicals (based on their MS/MS spectra) in >80% of the cases. This improved version called CFM-ID 3.0 is freely available as a web server. Its source code is also accessible online.

Metabolomic‐Based Approach to Identify Biomarkers of Apple Intake

“…Urine samples were further analyzed by LC‐MS/MS using a targeted approach with the same chromatographic conditions as described above and with three collision energies 10, 20, and 40 eV. MS/MS compound identification efforts included comparisons with authentic reference standards, fragmentation modeling using CFM ID, [ 46 ] and the HMDB [ 47 ] was used as a source. Metabolite Identification levels are reported according to Metabolomics standards initiative (MSI): with four levels of confidence in metabolite identification: Level I (Identified compounds); Level II (putatively annotated compounds); Level III (putatively characterized compound classes); Level IV (Unknowns).…”

Lipidomics from sample preparation to data analysis: a primer

“…Due to poor processing performance for large segmented molecules such as lipids, rulebased fragmentation was implemented in CFM-ID version 3.0 to predict the MS/MS spectra of 21 lipid classes, as well as a new scoring system and a chemical classification algorithm. These changes allow predicted ESI-MS/MS spectra of lipids to be obtained 200 times faster and ten times more accurately than when CFM-ID version 2.0 is used [110]. Thus, in-silico mass spectrometry is a fast-growing field, which should lead to advances in novel compound prediction and identification.…”

MS-CleanR: A feature-filtering approach to improve annotation rate in untargeted LC-MS based metabolomics

“…This may have important consequences by increasing both the false annotation rate and the number of ‘unknown’ features arising from wrongly attributed signals. This is especially true when the annotation process is based on in silico modeling of fragmentation patterns, as are Sirius 12 , MS-FINDER 13 , MetFrag 14 or CFM-ID 15 , since tandem mass spectrometry (MS/MS) spectra are processed without taking into account feature relationships. Thus, most untargeted metabolomics studies focus on a subset of identified metabolites for which spectral data are easily accessible from public repositories or in-house DBs.…”