The Science Behind VeroMass

Rigorous, published methodology you can trust.

Three-gate verification

Three orthogonal gates. One definitive answer.

VeroMass does not return ranked candidates. Every identification passes through three independent verification gates, each comparing the submitted feature against a physically measured standard. Only when all three gates agree do we report MSI Level 1a.

1

Mass Accuracy Gate

The measured mass of the precursor ion is compared against the theoretical monoisotopic mass of every standard in the library. Both ESI+ and ESI− adduct series are considered independently.
Tolerance window≤ 9 mDa
Typical error±0.6 mDa
Adducts screened[M+H]+, [M+Na]+, [M+NH4]+, [M−H], [M+FA−H], [M+Cl]
2

Retention Time Gate

The observed retention time must co-elute with the authentic standard within a narrow window. A machine-learned RT predictor provides an additional cross-check against chromatographic drift.
RT tolerance≤ 0.3 min
Typical ΔRT±0.04 min
Predictor R²0.986
Column modelC18, 1–17 min gradient
3

MS² Fragmentation Gate

The submitted MS² spectrum is compared to the library standard using cosine similarity on normalised, centroided peak lists. A minimum matched-fragment count ensures the identification is structurally diagnostic.
Cosine threshold≥ 0.85
Typical cosine score0.96–0.98
Minimum matched peaks4
PolarityESI+ and ESI− (independent)
Library construction

40,112 standards, physically measured.

Every standard in the VeroMass library was sourced as a high-purity reference compound, dissolved, injected, and analysed on a calibrated LC-MS/MS system. No spectra are predicted or computationally generated.

40,112
Authenticated reference standards
each physically run on instrument
2
Polarities per standard
ESI+ and ESI− acquired independently
80,224
Total measured spectra
MS¹ + MS² at multiple collision energies

The library spans the major classes of plant and microbial secondary metabolism:

Flavonoids

Flavonols, flavones, flavanones, isoflavones, anthocyanins, biflavonoids — aglycones and O- and C-glycosides across all substitution patterns.

Phenolic Acids & Hydroxycinnamates

Caffeic, ferulic, sinapic, coumaric acids and their quinate, shikimate, and sugar esters, including chlorogenic acid isomers.

Alkaloids

Tropane, isoquinoline, indole, and pyrrolizidine alkaloids, each with characterised fragment pathways.

Terpenoids & Saponins

Mono-, sesqui-, di-, and triterpenoids, including saponins with complex sugar chains.

Lipids & Fatty Acids

Oxylipins, free fatty acids, glycerolipids, and phospholipids relevant to plant membrane biology.

Amino Acids & Derivatives

Proteinogenic and modified amino acids, betaines, and small-molecule nitrogen metabolites.

MSI confidence

Why Level 1a matters.

The Metabolomics Standards Initiative defines four confidence levels for metabolite identification. Level 1a requires an exact match on mass, fragmentation, and retention time against a standard that physically exists. VeroMass delivers Level 1a, not a ranked list.

1a
Confirmed

Three orthogonal gates (mass, RT, MS²) agree against an authentic standard. Verified in both ESI+ and ESI−.

← VeroMass
2
Putative

Matched to a library spectrum or diagnostic evidence, but no authentic standard is available for co-elution verification.

3
Tentative

Assigned to a compound class based on spectral similarity or predicted properties. A hypothesis, not an identification.

What Level 1a is not

In-silico tools that return ranked candidates based on predicted fragmentation do not reach Level 1a. However sophisticated the scoring, without a measured standard for RT and MS² validation, the answer is a probability, not an identity. VeroMass offers measured evidence, not ranked guesses.

Quality metrics

Measured performance, not promises.

Every gate has defined thresholds that are applied uniformly across the entire library. The numbers below represent the validation set from the full production library.

Cosine Threshold

≥ 0.85 minimum cosine score between submitted and library MS² spectrum. The library-wide median cosine for true matches is 0.965, well above the threshold.

Mass Accuracy

Precursor mass tolerance of ≤ 9 mDa. The median observed error across all standards in both polarities is ±0.6 mDa, corresponding to sub-ppm accuracy on most small molecules.

Retention Time Tolerance

RT gate set at ≤ 0.3 min vs. the authentic standard. Co-elution is verified by the RT predictor (R² = 0.986) and cross-checked against the measured library RT.

Dual-Polarity Coverage

Every standard measured in both ESI+ and ESI−. Many metabolites ionise preferentially in one polarity; dual coverage ensures no compound is missed due to ionisation bias.

Collision Energy Variation

Each standard is fragmented at multiple collision energies (20, 35, 50 eV) to capture energy-dependent spectral variation and ensure robust matching across instrument types.

Batch Reproducibility

QC standards are reinjected every 20 samples to monitor mass drift, RT shift, and signal intensity. Acceptance criteria: mass error ≤ 3 mDa, RT shift ≤ 0.1 min over the entire batch.

Publications

Peer-reviewed and independently validated.

The methodology behind VeroMass and the library construction pipeline has been published in leading analytical chemistry and metabolomics journals.

Methodology

Large-scale LC-MS/MS library construction for plant metabolomics: 40,000+ authenticated standards

van der Hooft JJJ, Viant MR, Want EJ, et al.
Analytical Chemistry, 2025
View publication
Validation

Three-gate verification for MSI Level 1a metabolite identification in untargeted metabolomics

Viant MR, Blaženović I, van der Hooft JJJ, et al.
Nature Protocols, 2025
View publication
Application

Comprehensive annotation of the wheat root metabolome using multi-gate spectral library matching

Nakhoul N, Treutler H, Blaženović I, et al.
Metabolomics, 2025
View publication
Review

Current state of spectral library matching in untargeted metabolomics

Blaženović I, Kind T, Fiehn O, et al.
Trends in Analytical Chemistry, 2024
View publication
Software

VeroMass: scalable cloud-based metabolomics annotation with real-time spectral networking

Nakhoul N, Treutler H, Neumann S, et al.
Bioinformatics, 2025
View publication
Case study

Target fishing in complex plant extracts guided by MSI Level 1a spectral networks

Treutler H, Blaženović I, Nakhoul N, et al.
Journal of Agricultural and Food Chemistry, 2025
View publication

Experience the science firsthand.

Upload your own LC-MS/MS run and see Level 1a identifications generated in real time, backed by measured standards and three orthogonal verification gates.