Good practices and recommendations for using and benchmarking computational metabolomics metabolite annotation tools

Abstract

Background Untargeted metabolomics approaches based on mass spectrometry obtain comprehensive profiles of complex biological samples. However, on average only 10% of the molecules can be annotated. This low annotation rate hampers biochemical interpretation and effective comparison of metabolomics studies. Furthermore, de novo structural characterization of mass spectral data remains a complicated and time-intensive process. Recently, the field of computational metabolomics has gained traction and novel methods have started to enable large-scale and reliable metabolite annotation. Molecular networking and machine learning-based in-silico annotation tools have been shown to greatly assist metabolite characterization in diverse fields such as clinical metabolomics and natural product discovery. Aim of review We highlight recent advances in computational metabolite annotation workflows with a special focus on their evaluation and comparison with other tools. Whilst the progress is substantial and promising, we also argue that inconsistencies in benchmarking different tools hamper users from selecting the most appropriate and promising method for their research. We summarize benchmarking strategies of the different tools and outline several recommendations for benchmarking and comparing novel tools. Key scientific concepts of review This review focuses on recent advances in mass spectral library-based and machine learning-supported metabolite annotation workflows. We discuss large-scale library matching and analogue search, the current bloom of mass spectral similarity scores, and how molecular networking has changed the field. In addition, the potentials and challenges of machine learning-supported metabolite annotation workflows are highlighted. Overall, recent developments in computational metabolomics have started to fundamentally change metabolomics workflows, and we expect that as a community we will be able to overcome current method performance ambiguities and annotation bottlenecks.

Speciation of Nanocatalysts Using X-ray Absorption Spectroscopy Assisted by Machine Learning

“…(b) Embedding-based structural similarity in a low-dimensional in-silico database to find candidate structures and their distributions. Adapted with permission from ref Copyright 2022, The Author(s).…”

MS2Query: reliable and scalable MS2 mass spectra-based analogue search

“…However, all these approaches still have important limitations. Many of these methods were recently reviewed by our group, in particular those using machine learning 19 .…”

“…These methods show excellent results for smaller metabolites of <400 Da; however, for larger metabolites these approaches are still not fully reliable in returning correct elemental formulas and candidate structures. Besides that, the computation time to determine the fragmentation trees also increases substantially 19 . Natural mixtures typically contain considerable amounts of larger metabolites (>800 Da), and this thus poses challenges on the mass spectral interpretation.…”

“…This makes the modified cosine score less sensitive to a small chemical modification. However, multiple small chemical modifications can still result in a large decrease in mass spectral similarity based on the modified cosine score, which limits its ability to serve as a proxy for chemical similarity 19,[29][30][31] . Recently, two machine learning-based methods were developed that outperform cosine-based scores in predicting chemical similarities from MS 2 mass spectral pairs; the unsupervised Spec2Vec 30 and the supervised MS2Deepscore 32 .…”

Combining Experimental with Computational Infrared and Mass Spectra for High-Throughput Nontargeted Chemical Structure Identification

“…With the subsequent acquisition of MS/MS spectral data, a variety of computational methods are then used to score and rank candidate structures. , Computational options include: (1) fingerprint-based methods, , (2) in silico simulation methods, – and (3) combinatorial fragmentation methods. – However, because of the large number of candidate structures whose experimental exact masses fall within a given exact mass window, current GC/LC–MS/MS-based nontargeted metabolomics workflows seldom are able to identify more than 10–40% of detected compounds – unless annotation is limited to structural class or subclass only . Thus, the inclusion of additional orthogonal analytical and computational methods is needed in order to improve the likelihood that the correct structure is at the top of the ranked candidate list.…”