Building blocks for automated elucidation of metabolites: Machine learning methods for NMR prediction. Supplemental Material

These are additional figures and tables related to our dataset. The additional file input.csv.gz includes the input data as comma separated matrix, which can be read by most spreadsheet applications or the R code data <read.csv("input.csv", row.names=1). The first row is the atomID as stored in NMRshiftDB, the remaining 246 columns are the descriptor values with at most 50% NA values. The categorial descriptors were translated into discrete numeric values, and each descriptor was scaled up to 100.