ASICS

ASICS identifies and quantifies metabolites in proton nuclear magnetic resonance (¹H NMR) spectra from complex biological mixtures using statistical spectral decomposition against a library of reference metabolite spectra.

Key Features:

• Automated Metabolite Identification: Detects metabolites in complex ¹H NMR spectra using a curated library of pure metabolite reference spectra.
• Spectral Decomposition Model: Represents observed NMR spectra as a weighted linear combination of reference metabolite spectra.
• Chemical Shift Alignment: Applies a warping function to adjust reference spectra for local and global chemical shift distortions caused by experimental conditions such as pH, temperature, or ionic strength.
• Lasso-Based Metabolite Selection: Uses a lasso-type penalized estimator to perform sparse metabolite selection and robust quantification while resolving overlapping peaks.

Scientific Applications:

• NMR-Based Metabolomics Profiling: Quantifies metabolites in complex biological samples using ¹H NMR spectroscopy data.
• Metabolic Biomarker Discovery: Supports identification of metabolite signatures in biological fluids such as plasma or urine.
• Comparative Metabolomics Studies: Enables systematic analysis of metabolic differences across experimental or clinical sample groups.

Methodology:

ASICS models observed ¹H NMR spectra as a linear combination of reference metabolite spectra from a curated library, aligns spectra using a warping function to correct chemical shift variations, and estimates metabolite contributions using a lasso-type penalized regression framework.