VOC dashboard guide

Upload an Excel file with samples in columns, metadata rows on top, and a row named Compound or Compounds. The dashboard inspects VOC trends, replicate averages, and QC warnings.

1. Upload fileAccepted format: .xlsx / .xls.
2. Check QC firstOpen Sample overview and warnings to verify codes, days, and replicates.
3. Select compoundsChoose Top 10, All, or manually select VOCs.
4. Plot or exploreCompounds overview for the selected set; Explore VOC for single compounds or whole groups (individual / summed / z-scored).
Expected structure: Row 1: Sample names, e.g. 2.1.1A_Day 1, 2.1.1B_Day 1 Rows 2-x: Metadata, e.g. Day, Culture, Glucose, Replicate Compound row: Compound | sample 1 | sample 2 | ... Below: VOC peak areas

Or paste tabular data from Excel/Sheets. Tabs, semicolons, or commas are accepted.

Select compounds
Explore single compounds or whole VOC groups
Tip: set top-bar Colour by = Compound and Line style by = Culture or Substrate to compare selected VOCs across conditions.
AXIS
Aggregation — individual: one line per compound · summed: all compounds in the selected group/custom set are added within each replicate, then mean±error across A/B/C · summed + individual: thin top-N individual lines plus one bold all-compounds summed line. Recommended normalizations — Pareto / z-score: compare VOC patterns using log10(area+1) input · max: compare timing/shape; each compound peaks at 1 · fold-change: raw-area log2 change vs the matched Day0/protein-stock baseline (e.g. rec. casein uses rec. stock; native uses native stock; 0 = no change, +1 = 2×). For summed fold-change, the dashboard calculates log2(Σ raw sample area / Σ matched baseline area); compounds with a positive matched baseline stay in the denominator even if they are absent later, so disappearance can reduce the group. Lilly/variance scaling is kept for pathway overlays but can exaggerate very stable/noisy compounds. Axis — set Y scale/min/max/tick format independently. Replicates are averaged arithmetically first; global log10/sqrt/asinh display modes are applied only after the mean. When a normalization is active, Data and global Display/axis are disabled so hidden axis settings cannot change the result.
Heatmap controls
Colour encodes the data value (use the colour scale below). For a group strip use the Metadata ribbon. Top-bar colour/symbol/line controls don't apply here; heatmap has its own height.
Heatmap compounds
Heatmap uses the same selected compound set as the compound plots. Row-scaled modes (z-score / robust z / min-max / % max) scale each VOC separately so patterns are comparable despite different absolute areas. Robust z-score is less affected by one extreme peak.
PCA controls
For Scores/Trajectory, Colour by and Symbol by in the top bar group the PCA points. For Loadings/Contributors they are hidden because sample metadata cannot affect compound plots. The Top contributors control is only shown in the Top contributors view. Chart title, height and legend also come from the top bar.
Values sets what goes into the matrix; PCA scaling puts compounds on comparable footing — normally use both. For VOC pattern differences use log10(raw+1) + Pareto (or z-score); for total-intensity differences use raw/log with mean-center or none.
To approximate ClustVis (no transformation, unit-variance row scaling, SVD with imputation): set Values = raw, PCA scaling = z-score, Missing values = ClustVis-like SVD imputation, and turn Show baseline ON. Exact match is not guaranteed — PC signs/rotation can differ, and this dashboard uses a lightweight browser-side PCA.
VOC group composition
Bars are coloured by VOC chemical group. Chart height & legend come from the top bar; colour/line controls don't apply here.
Summarizes whether samples are dominated by ketones, alcohols, acids, sulfur compounds, etc. Group columns by one or multiple fields, e.g. Substrate + Culture + Day. Column labels only change the displayed text, not the calculation. Baseline / Day 0 samples are included when the baseline checkbox is on.
Fold-change ranking vs matched baseline
Ranks compounds by log2 fold-change against their matched baseline. For Exp3 this keeps recombinant stocks with recombinant samples and native stocks with native samples.
Interpretation: 0 = unchanged, +1 = 2× higher than matched baseline, +2 = 4× higher, −1 = 2× lower. Use this as a screening/ranking figure, then confirm key VOCs with raw/log trend plots.