MassFormer: Tandem Mass Spectrum Prediction with Graph Transformers

Young, Adamo; Wang, Bo; Röst, Hannes L.

doi:10.48550/arxiv.2111.04824

Cited by 9 publications

(20 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is therefore foreseeable that the efficiency of the interrogation of the 1M-DB will strongly benefit from up-to-date optimization of MatchMS parallelization as part of future version releases. Multiple tools have been developed in recent years to produce hypothetical MS/MS spectra (23)(24)(25)(26)49). A more recent development in this area is that of QCxMS, a computational tool that generate high-quality fragmentation spectra from molecules based on molecular dynamics calculations (26).…”

Section: Discussionmentioning

confidence: 99%

Evolutionary metabolomics of specialized metabolism diversification in the genus Nicotiana highlights N- acylnornicotine innovations

Elser,

Pflieger,

Villette

et al. 2023

Sci. Adv.

View full text Add to dashboard Cite

Specialized metabolite (SM) diversification is a core process to plants’ adaptation to diverse ecological niches. Here, we implemented a computational mass spectrometry–based metabolomics approach to exploring SM diversification in tissues of 20 species covering Nicotiana phylogenetics sections. To markedly increase metabolite annotation, we created a large in silico fragmentation database, comprising >1 million structures, and scripts for connecting class prediction to consensus substructures. Together, the approach provides an unprecedented cartography of SM diversity and section-specific innovations in this genus. As a case study and in combination with nuclear magnetic resonance and mass spectrometry imaging, we explored the distribution of N- acylnornicotines, alkaloids predicted to be specific to Repandae allopolyploids, and revealed their prevalence in the genus, albeit at much lower magnitude, as well as a greater structural diversity than previously thought. Together, the data integration approaches provided here should act as a resource for future research in plant SM evolution.

show abstract

Section: Discussionmentioning

confidence: 99%

Evolutionary metabolomics of specialized metabolism diversification in the genus Nicotiana highlights N- acylnornicotine innovations

Elser,

Pflieger,

Villette

et al. 2023

Sci. Adv.

View full text Add to dashboard Cite

show abstract

“…In comparison, it takes around 0.03, 0.05, and 24.1 seconds for NEIMS, MassFormer, and CFM-ID 4.0 to predict the MS/MS spectrum of a compound. We compare our model against the existing mass spectra prediction methods, including CFM-ID 4.0 (Wang et al, 2021), NEIMS (Wei et al, 2019) and MassFormer (Young et al, 2021). NEIMS and MassFormer are retrained by using our training set for a direct comparison.…”

Section: Performance Of Predicting Ms/ms Spectra Acquired By Using Ag...mentioning

confidence: 99%

“…The molecular point-sets encode the accurate 3D coordinates and attributes of the atoms, while the chemical bonds are represented in neighboring vectors. When trained and tested on the compounds' MS/MS spectra collected from the Agilent Personal Compound Database and Library (PCDL) and NIST20, 3DMolMS achieved higher accuracy than CFM-ID 4.0 (Wang et al, 2021), NEIMS (Wei et al, 2019) and the existing deep learning model MassFormer (Young et al, 2021). Furthermore, 3DMolMS achieves satisfactory performance when applied to the prediction of the spectra in the public spectral library MassBank of North America (MoNA) with minor fine-tuning, which indicates the 3DMolMS model can be generalized to the prediction of spectra acquired by different labs on different instruments.…”

Section: Introductionmentioning

confidence: 99%

“…PAINN makes use of not only the distances between atoms but also the angles between bonds. To the best of our knowledge, NEIMS (Wei et al, 2019) and MassFormer (Young et al, 2021) are the only existing neural network models for the compound MS/MS spectra prediction, which encode the small molecules as ECFP and molecular graphs, respectively, and then exploit the multilayer perceptron (MLP) neural network and Graph Transformers (Yun et al, 2019) for spectra prediction, respectively.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

3DMolMS: Prediction of Tandem Mass Spectra from Three Dimensional Molecular Conformations

Hong

Welch³

et al. 2023

Preprint

View full text Add to dashboard Cite

Motivation: Tandem mass spectrometry is an essential technology for characterizing chemical compounds at high sensitivity and throughput, and is commonly adopted in many fields. However, computational methods for automated compound identification from their MS/MS spectra are still limited, especially for novel compounds that have not been previously characterized. In recent years,in silicomethods were proposed to predict the MS/MS spectra of compounds, which can then be used to expand the reference spectral libraries for compound identification. However, these methods did not consider the compounds' three-dimensional (3D) conformations, and thus neglected critical structural information.Results: We present the3DMolecular Network forMassSpectra Prediction (3DMolMS), a deep neural network model to predict the MS/MS spectra of compounds from their 3D conformations. We evaluated the model on the experimental spectra collected in several spectral libraries. The results showed that 3DMolMS predicted the spectra with the average cosine similarity of 0.687 and 0.475 with the experimental MS/MS spectra acquired in positive and negative ion modes, respectively. Furthermore, 3DMolMS model can be generalized to the prediction of MS/MS spectra acquired by different labs on different instruments through minor fine-tuning on a small set of spectra. Finally, we demonstrate that themolecular representationlearned by 3DMolMS from MS/MS spectra prediction can be adapted to enhance the prediction of chemical properties such as the elution time (ET) in the liquid chromatography and the Collisional Cross Section (CCS) measured by ion mobility spectrometry (IMS), both of which are often used to improve compound identification.Contact: hatang@indiana.eduAvailability: The codes of 3DMolMS are available at <a href="https://github.com/JosieHong/3DMolMS">https://github.com/JosieHong/3DMolMS</a> and the web service is at <a href="https://spectrumprediction.gnps2.org">https://spectrumprediction.gnps2.org</a>.

show abstract

“…Representation learning of spectra such as Spec2Vec [28], MS2DeepScore [29], and sinusoidal embeddings [30] can be used to learn a more meaningful distance between spectra to facilitate molecular networking. Forward models have incorporated feed forward networks and graph neural networks to predict a fragmentation spectrum directly from molecular structure [31,32,33]. In the inverse direction MSGenie [34], Spec2Mol [35], and MetFID [36] directly attempt to generate fingerprints or SMILES strings [37] from mass spectra, but no approach outperforms CSI:FingerID when trained with equivalent data.…”

Section: Mainmentioning

confidence: 99%

Annotating metabolite mass spectra with domain-inspired chemical formula transformers

Goldman

Wohlwend²,

Stražar

et al. 2022

Preprint

View full text Add to dashboard Cite

Metabolomic studies have succeeded in identifying small molecule metabolites that mediate cell signaling, competition, and disease pathology in part due to large-scale community efforts to measure mass spectra for thousands of metabolite standards. Nevertheless, the vast majority of spectra observed in clinical samples cannot be unambiguously matched to known structures, suggesting powerful opportunities for further discoveries in the dark metabolome. Deep learning approaches to small molecule structure elucidation have surprisingly failed to rival classical statistical methods, which we hypothesize is due to the lack of in-domain knowledge incorporated into current neural network architectures. We introduce a new neural network driven workflow for untargeted metabolomics, Metabolite Inference with Spectrum Transformers (MIST), to annotate mass spectrometry peaks with chemical structures generalizing beyond known standards. Unlike other neural approaches, MIST incorporates domain insights into its architecture by forcing the network to more directly link peaks to physical atom representations, neutral losses, and chemical substructures. MIST outperforms both standard neural architectures and the state-of-the-art kernel method on fingerprint prediction from spectra for over 70% of metabolite standards and retrieves over 66% of metabolites with equal or improved accuracy, with 29% strictly better. We further demonstrate the utility of MIST in a prospective setting to identify new differentially abundant metabolite structures from an inflammatory bowel disease patient cohort and subsequently annotate dipeptides and alkaloid compounds without spectral standards.

show abstract

MassFormer: Tandem Mass Spectrum Prediction with Graph Transformers

Cited by 9 publications

References 54 publications

Evolutionary metabolomics of specialized metabolism diversification in the genus Nicotiana highlights N- acylnornicotine innovations

Evolutionary metabolomics of specialized metabolism diversification in the genus Nicotiana highlights N- acylnornicotine innovations

3DMolMS: Prediction of Tandem Mass Spectra from Three Dimensional Molecular Conformations

Annotating metabolite mass spectra with domain-inspired chemical formula transformers

Contact Info

Product

Resources

About