PLOS Computational Biology · 9 authors, 6 centres
This summary was generated by AI from a single paper. It has not been reviewed by a clinician and is not clinical advice. Verify against the source before acting on it.
SAMBA is a constraint-based method that uses random sampling in genome-scale metabolic networks to predict differentially exchanged metabolites under a metabolic perturbation. On 49 inborn errors of metabolism, SAMBA achieved 80% accuracy in predicting metabolite change direction, slightly exceeding the 76.8% accuracy of the FVA-based method.
SAMBA (SAMpling Biomarker Analysis) is a constraint-based modelling approach that simulates fluxes in exchange reactions of genome-scale metabolic networks following a metabolic perturbation using random sampling. It compares flux distributions between baseline and modulated conditions and ranks predicted differentially exchanged metabolites as potential biomarkers. The method was validated on 49 inborn errors of metabolism using 54 metabolites on Recon 2, achieving an accuracy score of 80% at a rank cut-off of 110, compared to 76.8% for the FVA-based method from prior work. A hypergeometric test confirmed that the number of experimental metabolites predicted in top ranks was statistically significant. Compared with FVA, sampling provides finer-grained z-score rankings rather than binary increase/decrease indicators, enabling intensity-based ordering of predictions. Limitations include computational cost, sensitivity to model accuracy, and difficulty in standard evaluation due to unmeasured metabolites potentially being misclassified as false positives.
