File formats

MAF / somatic mutation tables

Mutation landscape accepts MAF-style CSV files. Required columns:

Column	Purpose
Hugo_Symbol	Gene name
Tumor_Sample_Barcode	Sample identifier
Variant_Classification	Missense, nonsense, frameshift, splice site, silent, etc.
Chromosome	Chromosome label
Start_Position	Variant start coordinate
Reference_Allele	Reference base
Tumor_Seq_Allele2	Alternate allele

VCF files can be registered as mutation datasets; the pipeline reads MAF-style tabular exports for analysis.

Optional columns:

Column	Purpose
HGVSp_Short or Protein_Change	Protein change (e.g. p.V600E, p.Val600Glu) for the lollipop plot and the 3D structure view
TriNucleotide_Context or context	Trinucleotide context for SBS-96 signatures
Transcript_ID, ENSP	Transcript and protein identifiers; kept in the per-variant artifact for isoform resolution

Protein positions are assumed to refer to the UniProt canonical isoform. The 3D structure view validates each position's reference amino acid and lists mismatches rather than placing them.

Copy number alterations

Optional CSV for mutation landscape copy number summary:

Column	Required	Purpose
sample_id	Yes	Sample identifier
gene	Yes	Gene symbol
chromosome	No	Chromosome label
alteration	Yes	amplification or deletion

Clinical endpoints

Survival analysis reads clinical CSV files:

Column	Required	Purpose
patient_id	Yes	Patient identifier
time_to_event_days	Yes	Survival or PFS time in days
event_observed	Yes	1 = event, 0 = censored
treatment_arm	No	Treatment group for stratification

Molecular feature matrix

Optional CSV for survival stratification and Cox regression:

Column	Required	Purpose
patient_id	Yes	Join key to clinical table
Feature columns	No	TMB, immune scores, gene signatures, etc.

Longitudinal trajectories

Optional CSV for treatment timepoint analysis:

Column	Required	Purpose
patient_id	Yes	Patient identifier
timepoint	Yes	baseline, on_treatment, response, relapse
feature	Yes	Feature name
value	Yes	Numeric measurement

Per-cell expression (communication)

Column	Required	Purpose
cell_id	Yes	Cell identifier
Gene columns	Yes	Numeric expression values

Cell metadata (communication)

Column	Required	Purpose
cell_id	Yes	Cell identifier
cell_type	Yes	Annotated cell type label
condition	No	Treatment or response group for comparison

Ligand-receptor database (optional)

Custom CSV for cell-cell communication:

Column	Required	Purpose
ligand	Yes	Ligand gene symbol
receptor	Yes	Receptor gene symbol
pathway	No	Signaling pathway label

Bulk expression (immune profile)

Samples-as-rows, genes-as-columns CSV. First column should identify samples.

Repertoire (immune profile)

TCR/BCR clonotype table with clonotype identifiers, CDR3 sequences, and frequency counts. Format follows scirpy-compatible tabular exports.

Artifact outputs

Pipeline results under data/oncology/artifacts/{run_id}/:

Artifact	Contents
result.json	Full pipeline output with summary, tables, and visualization data

Single-cell and spatial omics

10x Genomics, Visium, Xenium, and bulk RNA-seq formats are handled by the Computational Biology area. Oncology links compbio datasets to study samples via shared patient/sample keys.

Whole-slide images

H&E and IHC whole-slide images are handled by the Pathology area. Oncology uses pathology outputs for spatial TME context without duplicating WSI ingestion.