ASICS: identification and quantification of metabolites in complex 1D 1H NMR spectra

1H Nuclear Magnetic Resonance (NMR) is a high-throughput technology that allows to obtain metabolomic profile from easy-to-obtain fluids (such as blood) at moderate cost. It is thus a promising tool to detect practically usable biomarkers. However, its interpretation can be hard to make, because metabolites present from the 1H NMR spectrum of a complex mixture are often identified and quantified only from expert knowledge. To facilitate the us e of such data, we developed a new R package, ASICS, that implements a method for the automatic identification and quantification of metabolites in 1H NMR spectra. The package combines all the steps of the analysis (management of a reference library with pure metabolite spectra, preprocessing, quantification, diagnosis tools to assess the quality of the quantification, post-quantification statistical analyses). The latest developments allow to improve alignment algorithm and quantification using information coming from all spectra of the same study. To assess the performance of ASICS, data from PORCINET (ANR-09-GENM-005) were used. Both the quantification and its impact on a post-quantification differential analysis were evaluated. Correlations between ASICS relative quantifications and biochemical dosages of three metabolites were computed a nd a similar analysis was performed with other quantification methods like Autofit or batman. These comparisons showed that ASICS allows for a faster and simpler direct biological interpretation than the classical bucket approach and obtains more precisely identified and quantified metabolites than other quantification methods. ASICS is released as an R/Bioconductor package.