CHA: A hepatocyte metabolomic atlas and connectivity map for Chinese herbs (processed metabolomics data)

This dataset contains the complete processed results of the Chinese Herbs Metabolic Perturbation Atlas (CHA), a quality-controlled resource of herb-induced metabolic responses in primary rat (Rattus norvegicus) hepatocytes. Primary rat hepatocytes were exposed for 24 h to decoctions of 161 traditional Chinese medicine herbs at 0.2 mg/mL (low dose) or 1 mg/mL (high dose), with three biological replicates per herb and dose and three untreated controls per batch. Metabolomes were profiled by untargeted LC-MS/MS (HILIC, BEH Amide, negative mode) in six analytical batches, paired with a parallel CCK-8 viability assay. A rule-based quality gate excluded batch 4 (median QC-RSD 43.4% > 40%; pooled-QC r = 0.825 < 0.85), retained for annotation only; the quantitative analyses cover 133 herbs. The five QC-passing batches (813 samples; median QC-RSD 20.0%–30.1%, pooled-QC r 0.901–0.989) were aligned and ComBat-corrected (batch F-statistic on PC1: 55,007.1 → < 0.01), yielding 15,414 feature groups (4,765 in ≥ 2 batches) across 266 herb–dose comparisons, with 2,622 annotation entries (81 MSI Level 2, 2,541 Level 3). Archives (tar.xz) contain, per batch: final peak tables, preprocessing QC, PCA/OPLS-DA, per-comparison differential statistics with volcano plots, metabolite annotation (MSI levels, m/z recalibration, SIRIUS/CSI:FingerID), KEGG pathway enrichment, heatmaps/boxplots, MetaboAnalystR advanced analyses (biomarker, chemical class enrichment, MSEA) and machine-readable web-export JSON. Cross-module archives provide the ComBat-corrected integrated matrices, CCK-8 toxicity classification, dose-response classes with a dose-response index, TCM category diagnostics, and the herb–herb signature similarity (connectivity map) analyses. Batch labels are encoded throughout: each herb was measured in a single batch and cross-herb contrasts are designed to be read within batch. The dataset powers the public CHA connectivity map at https://cha.qfxulab.com. See the included README.md for a full file inventory and usage notes. Licensed under CC BY 4.0; raw LC-MS/MS files are available from the corresponding author on request.