Table S4 ASV count data

These tab-separated files contain the same 18 biological samples: six each in Sham, OVX and OF44. Each ASV row includes a source feature identifier, 18 integer sequence counts, and the provider taxonomy string. The sample mapping is supplied in the sequencing QC table and sample_metadata.tsv.

ASV18_before_singleton_filter.tsv: provider non-chimeric ASV table before singleton filtering.
ASV18_filtered_before_rarefaction.tsv: filtered ASV table before rarefaction.
ASV18_after_rarefaction_39988.tsv: ASV table after sampling without replacement to 39,988 sequences per sample.
ASV18_depth_checks.tsv: column sums and numbers of detected nonzero ASVs at each stage.
sample_metadata.tsv: analysis sample identifiers, group assignments and sequencing sample identifiers.

The 39,988-sequence depth was a chosen normalization depth, not the lowest available filtered library depth (42,093). Some provider ASV matrix sums differ slightly from the provider processing report; both are retained with their stage definitions, without replacing one with the other. Feature rows with zero counts across this selected 18-sample subset are retained from the provider source and must not be counted as detected ASVs. No sample labels, taxonomy strings or count values have been changed.

The original raw read/read-pair counts, sample-level Q20/Q30, and complete DADA2 execution parameters were not available in the retrieved project archive. They have not been inferred from these count matrices. See Supplementary Methods and Table S3.
