RunSignatureAnalyzerOnFile: Run SignatureAnalyzer on a file containing a catalog AFTER...

In WuyangFF95/SynSigRun: Run Mutational Signature Analysis Software Packages Using Mutational Spectra generated by SynSigGen

Run SignatureAnalyzer on a file containing a catalog AFTER the SignatureAnalyzer code has been source'ed.

Description

Normally, please call SignatureAnalyzerOneRun instead of this function.

Usage

RunSignatureAnalyzerOnFile(
  input.catalog,
  out.dir,
  input.exposures = NULL,
  maxK = 30,
  tol = 1e-07,
  test.only = FALSE,
  delete.tmp.files = TRUE,
  overwrite = FALSE
)

Arguments

input.catalog	File containing input catalog. Columns are samples (tumors), rows are signatures. SignatureAnalyzer does not care about the row names (I think) TODO(Steve): check this.
out.dir	Directory that will be created for the output; abort if it already exits. Log files will be in paste0(out.dir, "/tmp").
input.exposures	A file with the synthetic exposures used to generate input.catalog; if provided here, this is copied over to the output directory for downstream analysis.
maxK	The maximum number of signatures to consider extracting.
tol	Controls when SignatureAnalyzer will terminate its search; tol was 1.e-05 for the PCAWG7 analysis.
test.only	If TRUE, only analyze the first 10 columns read in from input.catalog.
delete.tmp.files	If TRUE delete the many temporary files generated by SignatureAnalyzer.
overwrite	If TRUE, overwrite existing output

Details

Creates several files in out.dir:

1. sa.output.sigs.csv Normalized signatures (no all-0 signatures, column sums all 0)

2. sa.output.raw.exp.csv Raw exposures (attributions)

3. sa.output.exp.csv Same as sa.output.raw.exp.csv

4. sa.output.other.data.csv, contains a summary of important information, including the number of signatures extracted.

5. input.syn.exp.csv Optional, a copy of input.exposures, if it was provided.

Value

A list with the following elements:

1. signatures.W The raw signature matrix, *including* columns of all zeros.

2. exposures.H The raw exposure matrix, *excluding* rows of all zeros. The matrix product of the non-zero columns of signatures.w and exposures.H approximates the input spectrum matrix.

3. likelihood The likelihood as returned by SignatureAnalyzer.

4. evidence -1 * the posterior probability as returned by SignatureAnalyzer.

5. relevance One for each column of the signatures.W, as returned by SignatureAnalyzer.

6. error A measure of reconstruction error (?) as returned by SignatureAnalyzer

7. normalized.sigs The non-0 columns of signatures.W normalized so that each column sum is 1.

WuyangFF95/SynSigRun documentation built on Oct. 7, 2022, 1:16 p.m.