HS-SPME-GC-MS volatile profiling dataset of Bailan melon across harvest and short-term postharvest holding
Description
This dataset supports the manuscript “Early postharvest holding alters grafting-associated volatile profiles in Bailan melon”. It contains the processed HS-SPME-GC-MS volatile-feature matrix, standardized sample metadata, pooled-QC information, feature annotations, feature-level results for the five predefined comparisons, grafting-associated contrast-change results, representative volatile records, figure-source data, analysis code, software-environment information and regeneration instructions. Five sampling groups are represented: self-rooted fruit at harvest (45 days after pollination; group A), self-rooted fruit after 5 d holding (group B), self-rooted fruit after 10 d holding (group C), grafted fruit after 5 d holding (group D), and grafted fruit at harvest (group E). Each group contains three independent pooled composite samples, with each composite prepared from flesh of three fruit. The five comparisons are E vs A, D vs B, B vs A, D vs E and C vs B. The change in the grafted-to-self-rooted contrast over the matched 5-d interval is summarized as Delta_root = log2(D/B) - log2(E/A). Volatile-feature responses are internal-standard-normalized relative analytical responses rather than absolute concentrations. Compound names are treated as putative annotations unless authentic-standard confirmation is explicitly documented.
Files
Steps to reproduce
1. Load the processed volatile-feature abundance matrix and standardized sample metadata. 2. Match sample IDs and assign samples to groups A-E. 3. For PCA, transform relative responses as log2(x + 1e-9), then mean-center and scale each feature to unit variance. Calculate sample-level Pearson correlations from the same log2-transformed retained-feature matrix. 4. Apply the QC filter described in the manuscript: exclude features with pooled-QC coefficient of variation (QC-CV) >= 0.50. The retained matrix contains 1,047 volatile features. 5. Calculate group-mean relative responses for the five predefined comparisons: E vs A, D vs B, B vs A, D vs E and C vs B. 6. For each feature, calculate log2 fold change after adding a feature-specific pseudocount equal to one-fifth of the minimum positive study value. Use |log2FC| >= 1 to summarize large-magnitude changes. 7. Perform feature-wise Welch tests for the five comparisons and adjust p values using the Benjamini-Hochberg method. 8. Calculate Delta_root = log2(D/B) - log2(E/A) to describe the change in the grafted-to-self-rooted contrast between harvest and 5 d holding. 9. For the matched A/B/D/E subset, fit a feature-wise linear model to log2-transformed relative responses with grafting status, 5-d holding and their interaction; adjust interaction p values using the Benjamini-Hochberg method. 10. Recalculate Delta_root after omitting one pooled composite from each of A, B, D and E in all 3^4 = 81 combinations, and summarize sign retention and retention of |Delta_root| >= 1. 11. Use the deposited figure-source data and scripts to regenerate Figs. 1-4 and the associated summary outputs. 12. Treat compound annotations as putative unless authentic-standard confirmation is explicitly documented; do not interpret relative responses as absolute concentrations or odor activity values.