Figure 3 A-B source data files

low_density_table.csv - Precision and recall values for the top K ranked features (K = 1-40) identified by different methods using in silico subsampled microbiome data (< 1% dense). Used for panel A.

original_density_table.csv - Precision and recall values for the top K ranked features (K = 1-40) identified by different methods using the original microbiome data (50% dense). Used for panel B.

Notes:
- Precision and recall were calculated by comparing the top K ranked features against a curated set of known Microcoleus vaginatus-associated metabolites.
- Benchmarked methods include Joint-RPCA, MMvec, SPIEC-EASI, SparCC, and Spearman, Pearson, Phi, and Rho correlation metrics.