moiety_modeling

moiety_modeling deconvolves isotopologue profiles from mass spectrometry to derive moiety abundances of stable isotope-labeled metabolites and localize isotope incorporation for metabolic flux analysis.

Key Features:

• Moiety Model Representation: Decomposes metabolites into moiety units that represent metabolic transformations to structure isotopologue data.
• Model Optimization: Uses SAGA-optimize (simulated annealing and genetic algorithms) and additional optimization methods from the Python scipy library to solve boundary-value inverse problems for model parameter estimation.
• Model Selection: Supports model comparison and selection using Akaike Information Criterion (AIC) variants and Bayesian Information Criterion (BIC).
• Data Format: Encodes moiety models and associated isotopologue data in a JSONized format.

Scientific Applications:

• Stable isotope tracing: Analyzes isotopologue profiles from stable isotope labeling experiments measured by mass spectrometry.
• Metabolic pathway atom tracing: Enables tracing of individual atoms through metabolic pathways by localizing isotope incorporation across moieties.
• UDP-GlcNAc 13C timecourse analysis: Has been applied to deconvolute timecourse ^13C isotopologue data for uridine diphosphate N-acetyl-D-glucosamine (UDP-GlcNAc) in human prostate cancer LnCaP-LN3 cells.

Methodology:

Decompose metabolites into moiety units and model their metabolic transformations, perform parameter optimization using SAGA-optimize (simulated annealing and genetic algorithms) and scipy optimizers to solve boundary-value inverse problems, and apply AIC/BIC model selection criteria.