Mutation landscape

Mutation landscape analysis summarizes somatic variants across a patient cohort — TMB, variant classification, SBS-6/SBS-96 mutational signatures, oncoprint visualization, copy number alterations, driver pathway enrichment, and clinical-group mutation enrichment.

Prerequisites

MAF-style CSV with standard columns (Hugo_Symbol, Tumor_Sample_Barcode, Variant_Classification, etc.)
Optional trinucleotide context column (TriNucleotide_Context or context) for SBS-96 signature matrices
Optional copy number CSV with sample_id, gene, chromosome, and alteration columns, or GISTIC-like peak columns such as Cytoband, Wide Peak Limits, q values, and Genes in Wide Peak
Optional clinical metadata CSV with sample_id and a group column such as response_status, treatment_arm, tumor_type, or group_label

Launching the pipeline

Open the study Mutations page, select your MAF (and optional copy number / clinical) datasets, set the panel size and clinical group column, and click Run mutation landscape.

Key parameters

Parameter	Default	Purpose
`mutation_path`	—	Path to MAF-style CSV
`copy_number_path`	—	Copy number alteration or GISTIC-like peak table
`clinical_path`	—	Optional sample metadata for clinical enrichment
`clinical_group_column`	auto	Clinical group column for differential mutation analysis
`panel_size_mb`	38.0	Sequencing panel size for TMB per megabase
`top_genes`	20	Genes included in oncoprint matrix

Outputs

Output	Description
`sample_summary`	Per-sample variant counts and TMB per megabase
`gene_summary`	Gene-level mutation frequency across cohort
`oncoprint`	Sample × gene alteration matrix for visualization
`somatic_interactions`	Co-occurrence and mutual exclusivity pairs
`mutational_signatures`	Cohort and per-sample SBS-6 counts, optional SBS-96 matrix, COSMIC-like SBS assignment, cosine similarity, and clinical-group signature summaries
`copy_number`	Gene, chromosome, region/peak, and focal/arm CNA summaries; known driver CNA annotations; clinical-group CNA comparisons when sample-level CNA data are supplied
`pathway_enrichment`	Driver gene and oncogenic pathway enrichment
`clinical`	Clinical metadata columns and group summaries
`differential_mutation`	Two-group Fisher-test gene mutation enrichment with adjusted p-values
`clinical_pathway_enrichment`	Driver pathway enrichment by clinical group

The frontend Mutations page renders mutation summaries, top genes, driver pathways, dominant signature, SBS-96 context coverage, per-sample signature exposures, top CNA events, known driver CNA annotations, differential mutation hits, CNA clinical comparisons, and clinical pathway enrichment, alongside the interactive visualizations described below. Detailed signature and CNA matrices remain available through the run artifact payload.

Visualizations

Three genome-position-aware charts render directly on the page.

Oncoprint

Renders immediately from the run's oncoprint output. A gene × sample matrix (rows = the top mutated genes, columns = samples) with each cell colored by variant classification; genes are ordered by mutation frequency and samples are ordered to produce the familiar oncoprint "staircase." Clinical annotation tracks (response group, treatment arm, stage, etc.) render as colored strips beneath the matrix when a clinical dataset was supplied to the run, along with per-gene alteration percentages on the right edge.

Rainfall / kataegis plot

Plots one sample's mutations by genomic position (x-axis: chromosome) against the log10 inter-mutation distance (y-axis) — the standard "rainfall plot" for spotting localized hypermutation. A tight vertical cluster of points sitting low on the y-axis is a kataegis signal: many mutations packed into a small genomic window, usually from APOBEC activity. Detected kataegis loci (≥6 consecutive same-chromosome mutations with mean spacing ≤1kb) are called out in the chart header and marked in the plot. Because this needs every variant's coordinates rather than the run's aggregate summary, it lazily fetches a separate mutation_variants artifact the first time it's viewed. For very large variant counts (50,000+) you'll be asked to confirm before it downloads.

Mutation signature profile

A 96-substitution-class bar chart (the standard SBS-96 trinucleotide-context profile), colored by the six base substitution types, with a toggle to view any individual sample or the cohort-aggregate exposure. Renders directly from the run's mutational_signatures.sbs96_matrix output when trinucleotide context was supplied with the mutation file — no extra fetch or pipeline change needed beyond providing a TriNucleotide_Context (or context) column.

Pathway-level alteration burden

A horizontal bar chart of driver pathways (RTK/RAS, PI3K/AKT, P53, cell cycle, DNA repair, WNT, etc.) by cohort mutation frequency. When the run includes a clinical group comparison, each pathway instead shows two bars, one per clinical group, so a responder-vs-non-responder split reads at a glance; hovering a bar lists the specific driver genes mutated in that pathway and, when available, the enrichment q-value.

Gene detail panel

Select a gene from the dropdown (defaults to the most-mutated gene in the cohort) to see everything the run knows about it in one place: a known-cancer-gene badge (oncogene/tumor-suppressor role) and driver-pathway membership badges, summary mutation metrics, a lollipop plot of protein-position mutation clusters, copy-number-alteration evidence, and clinical/response association. All synced to the same gene selection.

The lollipop plot shows where mutations cluster along the gene's protein sequence: stems rise from a baseline to a colored dot at each mutation's amino-acid position, taller stems mean more samples/events at that exact residue, and the most-recurrent hotspot is labeled by name (e.g. "V600E"). Below it, a protein backbone bar shows curated domain annotations (sourced from UniProt/InterPro) so you can see whether hotspots fall inside a functional domain — BRAF's V600E, for example, sits inside its kinase domain. Domain annotations are currently curated for BRAF, TP53, KRAS, NRAS, PTEN, EGFR, PIK3CA, RB1, CDKN2A, BRCA1, BRCA2, MET, ALK, RET, MYC, TERT, MSH6, NF1, KIT, and ERBB2; other genes still plot correctly but show an estimated protein length and no domain track. (TERT's most common alterations are non-coding promoter mutations, which have no protein position and won't appear on this plot regardless of domain curation.) This shares a lazily-fetched mutation_variants artifact with the rainfall plot, and needs a parseable Protein_Change/HGVSp_Short column (standard HGVSp short form, e.g. p.V600E) to place points. Rows with missing or non-standard notation are silently excluded from the count.

Typical use cases

Rank driver alterations by frequency across a trial cohort
Compare TMB between responders and non-responders
Identify genes and driver pathways enriched in responders, non-responders, treatment arms, or tumor subtypes
Identify dominant mutational processes (APOBEC, HR deficiency, mismatch repair)
Stratify Survival analysis by TMB or specific driver mutations