The increasing size and complexity of mass spectrometry (MS) data sets necessitate advanced computational tools. This study presents Correland, a MATLAB-based software for clustering and visualizing metabolite correlations through weighted correlation networks, which directly represent pairwise associations. Its effectiveness was tested on a data set from nontargeted LC-MS analysis of 14-day-old Arabidopsis thaliana seedlings inoculated with Alternaria alternata and Fusarium oxysporum, demonstrating effective clustering of biosynthetically related metabolites. The ion grouping algorithm resulted in a substantial reduction in network scale (83 nodes/metabolites from approximately 900 features). In addition to network construction, Correland enables pseudomolecular ion identification with a success rate of 86-90% achieved in Arabidopsis extract. Network density is reduced by limiting visible edges, producing interpretable and visually coherent networks generated in a single step using rescaled Pearson correlation coefficients.
Kosinova et al. (Wed,) studied this question.