MetFusion

MetFusion integrates in silico fragmentation with spectral library and structural database information to improve compound identification from tandem mass spectrometry (MS/MS) data.

Key Features:

• In silico fragmentation (MetFrag): Uses the MetFrag algorithm to generate candidate structures and predict fragment ions from chemical structures.
• Spectral library integration: Matches experimental MS/MS spectra against curated spectral libraries such as MassBank and addresses limited coverage in collections like NIST.
• Structural database querying: Leverages structural databases including PubChem and ChemSpider by using predicted fragments to enable effective searches where direct spectral queries are absent.
• Score integration and re-ranking: Combines spectral library match scores with in silico fragmentation scores to re-rank candidate structures and improve identification accuracy.
• Empirical performance: Demonstrated on 1062 spectra, improving the rank of correct identifications from 28 with MetFrag alone to 7 using the combined approach.
• Scalability to larger databases: Methodology can be extended to larger chemical spaces such as KEGG.

Scientific Applications:

• Unknown compound identification: Identification of unknown metabolites and compounds from tandem mass spectra in metabolite profiling and screening experiments.
• Annotation across databases: Enabling annotation workflows that search both spectral libraries (MassBank, NIST) and large structural repositories (PubChem, ChemSpider, KEGG) when reference spectra are missing.
• Improved candidate ranking: Enhancing ranking accuracy in computational mass spectrometry pipelines for more reliable candidate selection.

Methodology:

Integrates MetFrag in silico fragmentation predictions with spectral library matching (MassBank) and uses predicted fragments to query structural databases (PubChem, ChemSpider), then combines scores to re-rank candidate structures.