Serum 1H NMR-based metabolomics of sole lesion development in Holstein cows

Supplementary Table 1. Details of the spectral bins selected as representative of each annotated metabolite and used in statistical analysis, including the Bovine Metabolome database identification number (BMDB ID).

Supplementary Table 2. Details of the unlabeled spectral bins (not annotated to a metabolite) used in the statistical analysis.

Supplementary Table 3. Results of univariable analysis with Wilcoxon signed-rank tests. Data were analyzed in 17 prespecified subsets split by time point, parity, and outcome definition; the control group in all cases were animals without sole lesions, outcomes were either cases of sole new hemorrhage (New SH), cases of new sole ulcers (New SU), those two groups combined (New SH/SU) or all cases of sole ulcers (All SU).

Supplementary Table 4. Variable selection stability results including the mean log (base 2) fold-change (Log2FC) and the baseline stability thresholds (T99 and T100, equivalent to an expected 1% and 0% false positive rate, respectively). Data were analyzed in 17 prespecified subsets split by time point, parity, and outcome definition; the control group in all cases were animals without sole lesions, outcomes were either cases of sole new hemorrhage (New SH), cases of new sole ulcers (New SU), those two groups combined (New SH/SU) or all cases of sole ulcers (All SU).