Input data

Every function names what it takes in the same words. A kind is one definition: what the data is, what the file has to look like, which record formats satisfy it. The Start board shows the kind beside each input with its explanation behind the question mark; a script reads the same list from sigscapeStudio.inputs().

A record's format is what the data panel decided when the file arrived (VCF, SBS-96 matrix, Exposure table, …). A kind can be met by several formats, and one format can serve several kinds: a VCF is SBS-96 samples for a spectrum plot, Somatic VCF for MuTopia, Mutations with read counts for Tau when it carries read support.

SBS-96 samplessbs96-samples

Mutation counts in the 96 trinucleotide classes, one spectrum per sample.

An SBS-96 matrix (channels like A[C>A]A down the first column, one column per sample, comma- or tab-separated), or a VCF or MAF of somatic SNVs with trinucleotide context, which the studio tallies into a spectrum.

Formats: SBS-96 matrix, VCF, MAF
Used by: Spectrum, De novo, Confidence, Map, Dossier, Compare, Artifacts, MuSiCal

Type	S1	S2
A[C>A]A	12	8
A[C>A]C	9	5
…

Signature matrix (W)signature-matrix

Reference signature profiles: 96 channels per signature, as COSMIC publishes them.

An SBS-96 matrix whose columns are signature names (SBS1, SBS3, …) and whose rows are the 96 channels; values are proportions or counts.

Formats: SBS-96 matrix
Used by: Compare, Diff, SigMA2

Type	SBS1	SBS3
A[C>A]A	0.0009	0.0203
…

Exposure table (H)exposure-table

How much of each signature each sample carries: a samples × signatures table of counts or fractions.

A CSV or TSV with sample ids in the first column and one column per signature (or the transpose). Extra text columns such as a tumor type are allowed and left out.

Formats: Exposure table
Used by: Compare, Crosswalk, SigMA2

sample,SBS1,SBS3,SBS5
SP99301,2030.3,0,4112.7
…

Add object JSONadd-object

A SigMA2 Add object: a cohort's exposures with the bins and clusters sigma2 annotated, and the simulations and models later stages wrote.

The JSON sigma2 saves with Add.save("….json") or the studio's own export: df, obs, bins, clusterInfo, metadata, signatures, samples; simulations and tuning when present.

Formats: Add object JSON
Used by: SigMA2

Tuning runtuning-run

What sigma2 tune() produced: one model per signature and panel, with its summary, folds, ROC and samples.

The folder tune(output_dir=…) wrote (tumor/assay/signature/training_summary.json and its tables), dropped as a folder; or an Add object JSON saved after tune().

Formats: SigMA tuning run, Add object JSON
Used by: SigMA2

Somatic VCFsomatic-vcf

One sample's somatic single-nucleotide variants with their positions.

A VCF (plain or gzipped) with CHROM, POS, REF and ALT; one sample per file. Positions must match the reference genome the tool is given.

Formats: VCF
Used by: MuTopia, Liftover

#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO
1	840113	.	T	C	.	PASS	…

Mutationsmutations

Per-mutation calls for a cohort: what sigma2 downsamples into panels when it builds an Add object.

A folder of VCFs in the runner's data folder (one file per sample, named by the sample or listed in the exposure table's path column), or one VCF or MAF.

Formats: VCF folder, VCF, MAF
Used by: SigMA2

Mutations with read countssnv-counts

Somatic SNVs with the tumor read support at each: what Tau turns into mutation multiplicity.

A VCF whose FORMAT carries AD (or INFO carries t_alt_count and t_ref_count), or an SNV table with columns chromosome, position, ref, alt, ref_counts, alt_counts (Tau's own table).

Formats: VCF, SNV table
Used by: Tau

chromosome	position	ref	alt	ref_counts	alt_counts
1	840113	T	C	19	9
…

Copy-number segmentscn-segments

Allele-specific copy number along the genome: the segments whose gains Tau times.

A table with chromosome, start, end, major (or nMajor) and minor (nMinor) copy number per segment; one row per segment.

Formats: CN segments
Used by: Tau

chromosome	start	end	major	minor
1	10001	121500000	2	1
…

Reference genome (FASTA)reference-fasta

The genome the variants are called against, for trinucleotide context.

A FASTA of the build the VCF uses, hg19 or hg38, with its .fai index beside it (on the runner, or dropped together), or a UCSC .2bit file. Optional for Tally: without one, the hosted hg19 and hg38 are used.

Formats: FASTA, 2bit
Used by: MuTopia, Explorer, Timing, Tally, Liftover, Tau

Variant callsvariant-calls

Somatic single-base substitutions with their positions, one sample or several.

A VCF (single- or multi-sample, plain or gzipped) or a MAF (one Tumor_Sample_Barcode per sample). Chromosome naming may be 1 or chr1.

Formats: VCF, MAF
Used by: Explorer, Timing, Tally

Chromosome	Start_Position	Reference_Allele	Tumor_Seq_Allele2	Tumor_Sample_Barcode
17	7578190	T	C	S1
…

De novo solutiondenovo-solution

Signatures extracted de novo, ideally at every number of signatures K the tool tried, with its stability and error statistics.

A SigProfilerExtractor output folder (dropped whole), a MuSiCal discovery run on the runner, or a signature matrix (channels down, signatures across).

Formats: SigProfilerExtractor output, SBS-96 matrix
Used by: De novo