Nebula

Processing...

Nebula

v0.699

Analysis Status

📊 Data matrix

Try example data

Copy & Paste

Upload data file

1. Choose file format

Row ID column for plots:

Intensity columns start at column 7; first 6 columns kept as annotations.

2. Choose file to upload

🔍 Loaded data matrix

📋 Meta table

Try example data

Copy & Paste

Upload meta file

Choose file to upload

Note: Meta table is auto-generated from matrix column names when you load a data matrix. Upload here to override.

Editable: Edit cells in the table on the right (Sample_ID is read-only).

Note: Meta table is auto-generated from matrix column names. Override by uploading in the sidebar.

Editable: Click cells to edit (Sample_ID is read-only)

Batch edit: Ctrl/Cmd+click to multi-select cells in one column, Shift+click for range, or use "Select All Visible (Same Column)".

Exclude samples

Outliers are pre-checked below. Uncheck to keep, or check additional samples, then apply.

Run outlier detection first.

Outlier score per sample

Outlier detection not run yet

Score every sample on 5 robust QC metrics (MAD-based); flag and exclude bad samples.

01Load a matrix in Data Preparation.

02Tune the flagging rule in the sidebar.

03Click Run outlier detection.

Filter Rows by Valid values

Scope

Minimum count of valid values Minimum percentage of valid values

Threshold

Column Select

Sel	Column

Pattern select

Pattern (substring / regex)

Contains Regex

Choose by groups

Group by (meta column) Then by (optional)

Choose a grouping column to list groups.

Filter columns by Valid values

Scope

Minimum count of valid values Minimum percentage of valid values

Threshold

No filtered data yet

Row Filter drops low-quality features (valid values, CV, keywords, sampling) before downstream analysis.

01Load a matrix in Data Preparation.

02Tune a criterion in the sidebar and click Apply.

03Review Passed / Filtered out / Summary here.

Annotation filters

Filter the annotation table below, then Apply to matrix. Loaded format: —. Only rows that pass all active criteria are kept for downstream analysis. You can also type into a column's header in the table to filter that column directly (included in Apply).

Phosphosite quick filters

Class I (unambiguous) sites only Residue

Map matrix row IDs to gene symbols, protein names, and UniProt accessions via MyGene.info (free). Results are stored for feature labels and Row Profile search (like DIANN annotations).

Species NCBI taxid (if Other) Interpret row IDs as

Idle.

API: MyGene.info (no key). Large tables are queried in batches; use Stop to keep partial results.

Offline: Serve over http://localhost if the browser blocks requests.

No name mapping yet

Map matrix row IDs to gene symbols, protein names and UniProt accessions via MyGene.info for feature labels and Row Profile search.

01Load a matrix in Data Preparation.

02Pick species and ID interpretation in the sidebar.

03Click Run mapping.

Annotation table: row order, sort, and pagination follow the Data Filter sub-tab.

Log

Overall ?

No matrix loaded yet

Overall dashboard summarizes the working matrix: per-column feature counts, intensity distributions and total signal.

01Load a matrix in Data Preparation.

02Return here and click Refresh dashboard.

Per-column summary

Horizontal D3 bars (one row per sample) using the same metric/transform rules as Data QC → Column correlation (sidebar). Y-axis shows shortened labels; each bar has an SVG tooltip with the full column id. Use Add to Report to embed the same chart as a PNG in the exported HTML.

Distribution across features (box plot)

Horizontal box plots per sample: log10(1 + I) for I > 0. Height scales with sample count; native tooltip shows full column name.

Total log signal per sample

Horizontal bars: Σ log10(1 + I) per sample over quantified features. Height scales with sample count.

Row Profile

Row label (list & plot legend)

Uses matrix row IDs if DIANN protein-group annotations are not loaded.

Quick selection

Top rows by intensity sum

Plot settings

Plot type Color theme Use log scale (Y axis) Z-score per row (across samples) — normalize each selected row to mean 0, sd 1 so trends are comparable between rows Mark contaminant proteins (Cont_) — flags them here and in the Column profile plots so per-sample contamination is easy to see

Tip: Search and select rows above, then click "Plot selected rows".

Row profile plot not generated yet

Plot intensity profiles of selected rows across all samples (bar, line or stacked composition).

01Load a matrix in Data Preparation.

02Search and select rows in the sidebar checklist.

03Click Plot selected rows.

Column profile ?

Display options

Sample column Row/feature label Max features (bar, curve, treemap, pie) Use log10(1 + intensity) for bar & line chart Mark contaminant proteins (Cont_) — red in bar/pie/treemap; shows % of sample intensity

No column profile yet

Per-sample views: ranked intensities, cumulative %, treemap, pie and log-intensity distribution.

01Load a matrix in Data Preparation.

02Pick a sample column in the sidebar.

03Click Refresh plots.

Column Correlation ?

Shared scale

Transform Treat 0 as missing

Pair of columns

Column X Column Y Feature label (hover) Bland–Altman

Matrices (subset)

Color theme Correlation metric Max columns in heatmaps

Distance from correlation

No column correlation yet

Sample–sample QC: correlation / distance heatmaps plus paired scatter, QQ and Bland–Altman views.

01Load a matrix in Data Preparation.

02Tune transform and metric in the sidebar.

03Click Refresh plots.

Scatter-correlation matrix (lower scatter / upper corr)

Distance matrix (subset)

Pair scatter (Y vs X)

QQ plot (sorted quantiles)

Bland–Altman

Correlation with column X

Data imputation is not recommended, missingness itself is informative, and imputation under a mismatched mechanism inflates type I error.

Imputation method

Sequential KNN parameters ?

K neighbors Distance metric Weighting Min. shared columns Fallback when no donor value Neighbor axis Max missing % per row (skip above) Group by (meta column) Max missing % per group per row (skip above)

Options

Value scale Empirical-Bayes shrinkage (ComBat)

No batch assessment yet

Detect batch effects across sample groups (silhouette, PCA) and correct them (median/mean centering, ComBat-lite).

01Load a matrix + meta table in Data Preparation.

02Pick the batch column in the sidebar.

03Review Detection, then Apply batch correction.

1. Phosphosite Matrix

Choose file format Remove Reverse & contaminants (MQ) Choose file to upload

2. Total Proteome Matrix

Choose file format Remove Reverse & Contaminants (MQ) Choose file to upload

3. Normalization Settings

Matching Columns

Phosphosite Match ID (e.g. Protein names): Proteome Match ID (e.g. Protein names):

Imputation (Sequential KNN)

Enable Sequential KNN imputation

K neighbors Distance metric Weighting Min. shared columns Fallback when no donor value Neighbor axis Max missing % per row (skip above)

Run Normalization from the sidebar to populate the normalized matrix — then click any row to plot its profile on the right.

Clustering

Cluster features (rows) Cluster samples (columns) Distance metric Linkage

Colors

Color scale Reverse scale (high = cold, low = hot)

Figure & margins

Title (optional)

Width (px)

Height (px)

Row dendro %

Column dendro %

Margin left (px)

Margin right (px)

Margin bottom (px)

Heatmap

Heatmap not generated yet

Hierarchical clustering of rows and columns with dendrograms, group annotations and optional row z-score.

01Load a matrix in Data Preparation.

02Tune clustering, sampling and colors in the left sidebar.

03Click Generate heatmap in the left sidebar.

Feature label (rows)

Applies to Cluster and visualize (current matrix) and to the heatmap-side Clustergrammer when a heatmap exists. Changing it refreshes an open Clustergrammer tab or rebuilds from the cached heatmap.

Distance Linkage Pre-filter rows by variance (optional) Log10 transform Z-score per row

Clustergrammer not loaded yet

Interactive clustered heatmap with dendrogram sliders, sharing the Heatmap tab clustering.

01Generate the Heatmap first.

02Clustergrammer loads automatically with the same clustered data.

PC to plot on X-axis: PC to plot on Y-axis: PC to plot on Z-axis (3D): Show Column Labels Auto label font/color Label font size (manual): Label color (manual): Auto show arrows Show arrows (manual) Show group confidence ellipses (95%) Plot Width (px): Plot Height (px): Main title (optional): Max Samples for PCA: Max Features for PCA:

Reducing features significantly speeds up PCA for large datasets

Number of Components: Fast Mode (fewer iterations, faster but less precise)

Performance: Using optimized typed arrays for faster computation. For best speed, reduce features to 200-500.

Data Preprocessing: Log10 Transform Column-wise Z-score Normalization Row-wise Z-score Normalization

PCA 2D

PCA plot not generated yet

Principal component analysis with group coloring, confidence ellipses and labels.

01Load a matrix in Data Preparation.

02Tune preprocessing and group annotations in the left sidebar.

03Click Generate PCA Plot in the left sidebar.

PCA 3D

PCA 3D plot not generated yet

Generate PCA to explore interactive 3D rotation, zoom, and metadata tooltips.

01Run Generate PCA Plot from the 2D Plot tab.

02Return here to rotate and zoom the 3D cloud.

PCA details

PCA details not available yet

Generate a PCA plot to see variance explained, per-sample scores, and top feature loadings.

01Run Generate PCA Plot from the 2D Plot tab.

02Return here for the scree plot, scores and loadings tables.

Distance metric: Coordinate on X-axis: Coordinate on Y-axis: Coordinate on Z-axis (3D): Show Column Labels Auto label font/color Label font size (manual): Label color (manual): Auto show arrows Show arrows (manual) Show group confidence ellipses (95%) Plot Width (px): Plot Height (px): Main title (optional): Max Samples for PCoA: Max Features for PCoA:

Distance matrix is samples × samples; reduce samples for very wide datasets.

Number of Coordinates: Data Preprocessing: Log10 Transform Column-wise Z-score Normalization Row-wise Z-score Normalization

PCoA 2D

PCoA plot not generated yet

Classical multidimensional scaling on sample distances (Bray-Curtis, Euclidean, Manhattan, Cosine, Correlation).

01Load a matrix in Data Preparation.

02Pick a distance metric in the left sidebar.

03Click Generate PCoA Plot in the left sidebar.

PCoA 3D

PCoA 3D plot not generated yet

Generate PCoA to explore interactive 3D rotation, zoom, and metadata tooltips.

01Run Generate PCoA Plot from the 2D Plot tab.

02Return here to rotate and zoom the 3D cloud.

Perplexity (5–50): Max iterations: Show Column Labels Auto label font/color Label font size (manual): Label color (manual): Auto show arrows Show arrows (manual) Plot Width (px): Plot Height (px): Main title (optional): Max Samples: Max Features:

t-SNE works well for many samples; limit features for speed.

Data Preprocessing: Log10 Transform Column-wise Z-score Normalization Row-wise Z-score Normalization

t-SNE plot not generated yet

Non-linear neighbor embedding of samples with group coloring and labels.

01Load a matrix in Data Preparation.

02Tune perplexity and preprocessing in the left sidebar.

03Click Generate t-SNE Plot in the left sidebar.

Non-linear neighbor embedding that preserves both local and global structure (McInnes et al., 2018). Powered by umap-js (JS port of umap-learn), run here in the browser.

Neighbors (n_neighbors, 2–100): Min. distance (min_dist, 0–0.99): Spread: Epochs (blank = auto): Random seed: Show Column Labels Auto label font/color Label font size (manual): Label color (manual): Auto show arrows Show arrows (manual) Plot Width (px): Plot Height (px): Main title (optional): Max Samples: Max Features:

UMAP scales well to many samples; limit features for speed. Standardize features (column z-score) so no feature dominates the neighborhood graph.

Data Preprocessing: Log10 Transform Column-wise Z-score Normalization Row-wise Z-score Normalization

UMAP

UMAP plot not generated yet

Non-linear embedding preserving local and global structure (McInnes et al., 2018).

01Load a matrix in Data Preparation.

02Tune neighbors and preprocessing in the left sidebar.

03Click Generate UMAP Plot in the left sidebar.

Groups samples (columns) by their feature profiles — an exploratory way to reveal potential groups when no metadata grouping exists.

Choose number of clusters (k): Selection metric (auto):

k min:

k max:

Fixed k (when mode = Fixed): Algorithm: Initialization: Restarts (n_init): Max iterations: Random seed: Sampling (speed): Max samples: Max features:

Silhouette analysis is skipped above 2000 samples (it grows with the square of the sample count).

Data Preprocessing: Log10 Transform Column-wise Z-score (per feature) Row-wise Z-score (per sample)

Standardizing features (column z-score) is recommended so k-means is not dominated by high-abundance features.

Display: Show centroids Show sample labels Plot width (px): Plot height (px): Main title (optional):

Clusters (PCA projection)

K-means not run yet

Cluster samples with automatic k selection (silhouette / elbow), projected on PCA.

01Load a matrix in Data Preparation.

02Choose fixed k or auto selection in the left sidebar.

03Click Run K-means Clustering in the left sidebar.

Model selection (how k was chosen)

Elbow — WCSS vs k

Silhouette — mean vs k

Cluster assignments

UpSet / Venn / Karnaugh

Presence rule

Present if value is a finite non-zero number (0 counts as missing)

Present if intensity is strictly greater than a threshold (low values = low confidence; 0 still counts as missing):

Threshold Default 0 ⇒ same as >0

Set combinations (UpSet plot)

Ordering

Mode

Minimum set members (degree)

Maximum set members (degree)

Max # combinations (after sort)

Include empty combinations

Help & about this tab

Each set is one sample column. A row (feature) belongs to a set if that cell passes the presence rule above. Three linked views (see UpSet.js components) use UpSet.js (AGPL-3.0). Each chart’s Pop out button (next to its title in the main panel) opens popout/upset_popout.html via postMessage (needed for file://); allow pop-ups. The Heatmap tab uses the same pattern with the shared popout/plot_popout.html. Plot toolbar PNG/SVG/dump/VEGA stay on the default UpSet.js controls.

No set intersections yet

Each set is one sample column; linked UpSet, Venn and Karnaugh views show feature overlaps by presence rule.

01Load a matrix in Data Preparation.

02Tick sample columns in the set chooser.

03Click Refresh plots.

Hover a region in any view to highlight the same intersection in the others (linked selection).

UpSet

Venn diagram

Karnaugh map

Proportional Venn

Shape

Color theme

Notes & attribution

Circle areas are proportional to each set's feature count; the overlap between two circles is placed to match the intersection count (exact for 2 sets, approximate for 3). For 4+ sets the whole layout is fitted area-proportionally by the Eunoia library loaded from a CDN (falls back to sized circles when offline). The Shape dropdown above picks the shape family used by the fitter — Auto (circles for 1–3 sets, ellipses for 4+), Circle, Ellipse, Square, or Rectangle: with an explicit shape, 1–3 sets are fitted by Eunoia too (or the built-in solver if the CDN is unreachable), and a single selected set draws that shape directly. The Color theme dropdown changes the set/region palette. Hover a shape or region for the exact numbers.

Comparison

Comparison mode Group by (meta column)

Group A (Usually "Baseline" group) Group B (Usually "Treatment" group)

Significance cutoffs

|log2FC|

Fold change

p-value cutoff

FDR cutoff (when BH)

Data & test

Feature label (table & hover)

Log2(x+1) 0 / invalid = missing

Min valid / group

FC pseudocount

Test (two groups) Multiple testing

Plots & display

Volcano Y-axis Show labels for significantly changed

No Differential results yet

Compare sample groups on the working matrix joined to the meta table — two-group tests (Welch / Student / limma moderated t) or multi-group ANOVA / Kruskal–Wallis — with volcano, MA, p-value histogram, group heatmap and a sortable results table.

01Load matrix + meta in Data Preparation → Data PreProcess.

02In the left sidebar, pick Group by + groups and tune the cutoffs.

03Click Run differential analysis — plots and the results table appear.

Top N features by raw p-value; matrix shows group means on the tested scale (optional row z-score). Run multi-group analysis first.

Showing only Significant Changed Show sample quantification columns

Feature	mean A	mean B	log2FC	t	df	p	FDR	sig

Detail plot: Click a row to plot its per-sample / per-group quantification.

Sample grouping

Use Status column as-is (values T / C)

Group by (meta column) Control group (→ C) Treatment group (→ T)

Bait column (required)

Input settings

Input level Protein column name Peptide column name Fragment column name

No SAINT results yet

Score bait–prey interactions (AP-MS) from the matrix + meta table with control / treatment mapping and a Bait column, then explore the network and scatter views.

01Load matrix + meta in Data Preparation → Data PreProcess.

02In the left sidebar, map Group by → control / treatment and choose the Bait column.

03Click Run SAINT analysis — the results table, network and scatter appear.

Open this tab to load libraries.

No enrichment results yet

Test a gene list for term enrichment against Enrichr libraries — paste genes, or reuse matrix row labels, Differential results, or SAINT preys — with results table, enrichment plots and heatmap.

01Pick Species + Category / Library in the left sidebar.

02Choose the gene list source (paste, matrix, Differential, SAINT).

03Click Run enrichment — the results table and plots appear.

Plot options

Uses the full Enrichr term list (not only the current results page). Pick a view below; use Redraw plots after changing options.

Results heatmap

Same clustering pipeline and layout as the top-level Heatmap tab, but the plotted matrix comes from the Enrichr Results table: rows = ontology / library terms, columns = per-sample intensities attached to those terms (not the original protein × sample matrix). Statistics below only filter which terms are included.

Row filters

Leave a field empty to skip that filter.

Max adj. P-value Max P-value Min combined score Min odds ratio Min overlap (genes)

Clustering

Cluster ontology terms (rows) Cluster samples (columns) Distance metric Linkage

Colors

Color scale Reverse scale

Figure & dendrogram %

Title (optional)

Width px

Height px

Row dendro % Column dendro % Margin L Margin R Margin B

From the Results sub-tab, run enrichment with a loaded matrix so each term has per-sample intensities. Adjust filters, then Generate clustered heatmap (ontology terms × samples). Group bars use the same meta checkboxes as the Heatmap tab when columns match Sample_ID.

Ranked gene list

Differential results Expression matrix (direct, no Differential run) Paste list Upload file (.rnk/.tsv/.csv)

Ranking metric

ANOVA results rank by -log10(p) (unsigned). Duplicate symbols collapse to max |score|.

Parameters

Weight p Permutations Min set size Max set size Seed

No GSEA results yet

Preranked gene-set enrichment (Subramanian et al. 2005) over MSigDB collections — rank genes from Differential results, the expression matrix, a pasted list or a file, then test whole collections at once.

01Pick the ranked gene list source in the left sidebar.

02Tick MSigDB collections / branches (or upload a custom .gmt).

03Click Run GSEA — the results table and enrichment plot appear.

Method

gsva (classic KS walk) ssgsea (Barbie et al.) plage (Tomfohr et al., SVD eigengene) zscore (Lee et al., combined z-score)

Weight tau maxDiff (max+min, default) absRanking (Kuiper, maxDiff only) CDF kernel (kcdf)

ECDF min sample size (auto only)

No GSVA results yet

Per-sample pathway scores (Hanzelmann et al. 2013) — gsva, ssgsea, plage or zscore — over MSigDB collections, shown as a score heatmap plus a scores table.

01Pick the method and matrix options in the left sidebar.

02Tick MSigDB collections / branches (or upload a custom .gmt).

03Click Run GSVA — the score heatmap and scores table appear.

Score heatmap

01 · Analysis

Group by (meta column) Feature-group column (rows → groups)

Group A

Group B

Target FDR

Bootstrap B

null-probability model (multi-feature groups)

Intensity centering

Intensity: missing values

Intensity: test statistic

About Hybrid Differential

Combines a presence/absence test (exact binomial for single-feature groups, parametric bootstrap for multi-peptide groups) with an intensity test (limma moderated t / Welch) into one feature list at a single overall FDR. Excels for one-state proteins — observed in one group, absent in the other — where intensity alone has no contrast.

Use when missingness itself is informative (MNAR), or vs. plain Differential for presence-driven hits
Input is the current Nebula matrix + meta table; pick any meta column with n ≥ 2 per group

Wang et al. (2012) Bioinformatics 28:1586–1591. doi:10.1093/bioinformatics/bts193

Nothing analyzed yet

The Hybrid Differential test (Wang et al. 2012) merges two views of the same matrix: Stage I — presence/absence (exact binomial for single-feature groups; bootstrap for multi-peptide groups) and Stage II — intensity (log₂, optional median centering / min−1.5·IQR imputation, limma moderated t or Welch). Stage III builds one list at a single hybrid FDR Σ w·[cp+(1−cp)F_bin]/ #selected. It is the complement to plain Differential: use it when detection itself is differential (one-state proteins).

01Load & preprocess in Data Preparation → Data PreProcess (matrix + meta; Nebula is the sole data source).

02In the left sidebar, pick Group by + Group A / B (needs n ≥ 2 per group) and tune Target FDR / test.

03Click Run hybrid analysis — Feature table, Dataset, Analysis, Volcano, Hybrid scatter and Diagnostics appear.

How to read: H = hybrid (primary) list · B = presence/absence list · I = intensity list (all at same cp*); pep = per-feature posterior, fdr = list-level at min p. One-state badge = observed in one group only (paper's flagship case).
Reference: Wang et al. Bioinformatics 28(12):1586–1591 (2012) · doi:10.1093/bioinformatics/bts193.

Feature table

click a header to sort

pMin = min(pBin, pInt); “sel” columns mark membership in each selection at the target FDR. FDR estimates are per the paper’s method; see Diagnostics for self-tests and notes.

Dataset

—

Stage I — presence/absence

paper §2.3–2.5

Per feature group, presence counts in the two groups are tested for differential presence (exact binomial for single-feature groups; parametric bootstrap for multi-feature groups), and an FDR curve is estimated from the mixture of null and alternative.

Fig.2 p-hist

Fig. 2aDistribution of presence/absence p-values (log₁₀ scale, dashed = uniform null).

Fig. 2bNull-probability π₀ estimate vs p-value cutoff (quadratic fit).

Stage II — intensity

Storey–Tibshirani FDR

Quantities are log₂-transformed, (optionally) median-centered, left-censored small values are imputed, and a Welch t-test per feature group yields p-values; π₀ and the FDR curve follow Storey & Tibshirani (2003).

Stage III — hybrid selection

paper §2.6 · single overall FDR

The two stages are combined into one selection rule with a single false-discovery-rate estimate (hybrid FDR ≥ component FDRs by construction). Selected features are the union of the two component selections at their operating cutoffs.

Fig.1 x-axis

Fig. 1Number of selected features vs estimated FDR for the three analyses; vertical lines mark the operating points at the target FDR.

One-state Fig.4 — highlight

Fig. 4One-state: presence ratio vs intensity FC correlation (Pearson r / Spearman ρ) and same points in both coordinates. Needs impute=on for intensity axis.

Volcano plot

fold change vs −log10 p

Effect size vs significance — colours answer which method found it. Modes: significance (x = quantitative log2FC vs y = qualitative Δ presence), hybrid (union list coloured by origin), presence/absence, intensity. Dashed lines mark target FDR and |FC|=1.

Fig. 3Volcano / significance plots — x = log₂FC, y = −log₁₀ p; colours show origin of selection.

Hybrid scatter

one.png style

Each point is one feature group — x = quantitative log₂FC, y = qualitative Δ presence. Colour is the two-by-two significance table at the target FDR. Dashed lines mark |log₂FC|=1, |ΔP|=0.3 and ΔP=0, identical to one.png.

Fig. HSHybrid scatter (quantitative vs qualitative) — the one.png style figure rebuilt from the Feature table.

Diagnostics & method notes

self-tests + deviations

Method (as in the paper, hybrid2 port)

For two comparison groups with n1, n2 samples, a feature is “present” when observed (>0). Stage I: exact binomial for single-feature groups, parametric bootstrap (structured/unstructured) for multi-feature groups. Stage II (hybrid2): log₂ → optional per-sample median centering (off by default, assumes null majority) → optional left-censored imputation at min −1.5·IQR (off by default: complete case, one-state features get pInt=— and are covered by Stage I) → by default limma moderated t (empirical-Bayes, prior df₀,s²₀ via fitFDist, borrowed across features, fractional df via regIncBeta; Welch selectable). Stage III: weighted hybrid FDR Σ w·[cp+(1−cp)F_bin]/ #selected, one list. Table adds pep (local posterior L₀/(L₀+L₁), flat 50/50) and fdr (global list-level FDR at min p, Fig.1 axis, capped at 1).

Deviations (hybrid2)

Intensity: log₂ + (optional median centering, off) + (optional imputation min−1.5·IQR, off) + limma moderated t (default, verified vs R limma 3.58.1 to 1e-12) or Welch; Storey–Tibshirani π₀ (quadratic fit to π₀(λ) extrapolated to λ→1). Exact exactTails now cumulative (bugfix), digamma/trigamma for limma prior, per-feature stdev.unscaled uses observed n, explicit 0/NA handling. Bootstrap weight f₀/f₁ adaptive window δ=0.01 doubled if empty, 2-sided |T*|. See hybrid2 README.

Notes

Input data are assumed pre-filtered (contaminants/decoys removed). Two groups are required; choose any two levels of the meta column. Features with no observations in either group contribute no signal.

Preview builds when you open the Report tab (with a matrix loaded) or when you change section checkboxes.

Data Source

No file loaded

No fasta loaded

No XIC data

Filtering Parameters

Precursor Q.Value ≤ 0.010

Global Q.Value ≤ 0.010

Protein Group Q.Value ≤ 0.010

Quantification Settings

Intensity Metric

Protein Matrix Quantification

Proteotypic Peptides Only

Dataset File

Precursors (Filtered / Total)

0 / 0

Protein Groups

Unique Peptides

Sample Runs

Per-sample identifications (current filters)

Peptide counts

Click any protein row to inspect its peptides in the Peptide table

⏳

No protein data loaded

Please load a DIA-NN parquet file to view protein quantification.

Showing 0 of 0 proteins

Rows: Page 1

Peptide counts

Genes aggregated from protein groups — click a row to filter

🧫

No gene data loaded

Please load a DIA-NN parquet file to view gene quantification.

Showing 0 of 0 genes

Rows: Page 1

Target Protein: None selected (showing all peptides)

💬

No peptides found

Select a protein in the Protein Table or adjust your FDR filtering parameters.

Showing 0 of 0 peptides

Rows: Page 1

🔭

No spectra to explore

Load a dataset to browse proteins, peptides and runs.

Protein

Precursor

Run

y b ms1 theoretical

No protein selected

Select a protein from the list to map its peptides onto the protein sequence.

🔍 Loaded data matrix

Export meta table

Batch Edit Selected Cells

Hierarchy columns

Outlier score per sample

Outlier detection not run yet

Robust z-scores (samples × metrics)

Per-sample metrics and flags

No filtered data yet

No filtered data yet

No name mapping yet

No matrix loaded yet

Per-column summary

Distribution across features (box plot)

Total log signal per sample

Row Profile

Quick selection

Plot settings

Row profile plot not generated yet

Column profile ?

Display options

No column profile yet

Treemap (top N + Other)

Ranked intensity (bars + connecting line)

Cumulative % of total (sorted order)

Pie chart (top N + Other)

Intensity distribution (all features, log10 scale)

Column Correlation ?

Shared scale

Pair of columns

Matrices (subset)

No column correlation yet

Scatter-correlation matrix (lower scatter / upper corr)

Distance matrix (subset)

Pair scatter (Y vs X)

QQ plot (sorted quantiles)

Bland–Altman

Correlation with column X

No batch assessment yet

Clustering

Colors

Sampling

Preprocessing

Row labels

Sample group overlays ?

Heatmap

Heatmap not generated yet

Clustergrammer not loaded yet

Group Annotation Labels

PCA 2D

PCA plot not generated yet

PCA 3D

PCA 3D plot not generated yet

PCA details

PCA details not available yet

Variance explained (scree)

Sample scores (PC coordinates)

Top feature loadings

Group Annotation Labels

PCoA 2D

PCoA plot not generated yet

PCoA 3D

PCoA 3D plot not generated yet

Group Annotation Labels

t-SNE plot not generated yet

Group Annotation Labels

UMAP

UMAP plot not generated yet

Clusters (PCA projection)

K-means not run yet

Model selection (how k was chosen)

Elbow — WCSS vs k

Silhouette — mean vs k

Cluster assignments

UpSet / Venn / Karnaugh

Presence rule

No set intersections yet

UpSet

Venn diagram

Karnaugh map