Metchalizer

Metchalizer normalizes untargeted metabolomics data by removing batch effects to enable accurate detection of metabolite biomarkers for screening and diagnosis of inborn errors of metabolism (IEM).

Key Features:

• Normalization methodology: Employs 17 stable isotope-labeled internal standards combined with a mixed effect model to reduce batch effects across multiple independently processed batches and enable integration of out-of-batch controls.
• Covariate regression: Incorporates a regression model that accounts for age- and sex-related covariates fitted on control samples from all eight analyzed batches to adjust demographic variation in metabolite levels.
• Log-transformed processing: Applies log-transformation of data to enhance mitigation of batch effects.
• Performance evaluation: Was evaluated against existing normalization methods using six different metrics to assess batch effect removal and biomarker detection.
• Biomarker detection: Demonstrated identification of 178 known biomarkers across 45 IEM patient samples during comparative evaluation.
• Age-dependent variation detection: Revealed that 10–24% of metabolomic features exhibited significant age-dependent variations.
• Clinical correction scope: Enables establishment of reference values and correction for variations due to diet, age, gender, and technical batch differences in semi-quantitative untargeted metabolomics.

Scientific Applications:

• IEM diagnosis: Applied in laboratory settings to detect metabolite biomarkers for diagnosing inborn errors of metabolism.
• Reference value establishment: Enables use of large-scale out-of-batch control samples to establish clinically relevant reference values for metabolite concentrations.
• Clinical metabolomics research: Supports clinical metabolomics studies by correcting for batch, demographic, and dietary variation to increase throughput and detection accuracy.

Methodology:

Integrates 17 stable isotope-labeled internal standards, applies log-transformation, uses a mixed effect model for normalization, fits a regression model with age- and sex-related covariates on control samples from eight batches, and evaluates performance using six metrics.