CliqueMS

CliqueMS annotates in-source metabolite ions from LC-MS untargeted metabolomics data by grouping coeluting features and inferring parental masses to support metabolite identification.

Key Features:

• Coelution Profile Similarity: Leverages similarity between coelution profiles to associate features that originate from the same metabolite.
• Discriminatory Feature Similarity Metric: Employs a highly discriminatory metric to assess feature similarity and differentiate closely related ion signals.
• Generative Model: Incorporates a simple generative model to transparently represent similarities between features.
• Maximum Likelihood Inference: Uses maximum likelihood inference to group features derived from the same metabolite.
• Empirical Adduct Frequencies: Utilizes empirical adduct frequency data to identify parental masses from observed ions.
• Flexible Parental Mass Identification: Proposes and ranks alternative parental mass annotations to explore multiple candidate identifications.

Scientific Applications:

• Complexity Reduction: Condenses thousands of LC-MS features into a reduced set of metabolites to simplify downstream analysis.
• Validation on Standards and Biological Samples: Validated using simple mixtures of standards and complex biological samples.
• Comparative Annotation Performance: Annotates metabolites and adducts from single spectra with higher correctness relative to existing tools as reported.
• Metabolomics and Translational Research: Supports determination of metabolite numbers and identities and the discovery of metabolic pathways, biomarkers, or therapeutic targets in pharmacology, toxicology, and systems biology.

Methodology:

Computational steps explicitly include leveraging coelution profile similarity, applying a discriminatory feature-similarity metric, modeling feature similarities with a simple generative model, using maximum likelihood inference to group features, employing empirical adduct frequencies to identify parental masses, and proposing and ranking alternative annotations.