SynSigEval: SynSigEval

In WuyangFF95/SynSigEval: Evaluate Results of Mutational Signature Analysis Software on Synthetic Spectra Created by Package SynSigGen

SynSigEval

Description

Assess the performance of two steps in mutational signature analysis:

• signature extraction

• exposure inference (a.k.a. signature attribution)

by computational approaches, using catalogs of synthetic mutational spectra created by package SynSigGen.

Input

SynSigEval requires the input data listed below:

1. E, matrix of synthetic exposures (signatures x samples)

2. S, mutational signature profiles (mutation type x signature)

3. synthetic.spectra, synthetic mutational spectra with known ground-truth mutational signature profiles (S) and exposures (synthetic.exposures). It can be created from SynSigGen.

4. T, signatures extracted by SignatureAnalyzer, SigProfiler, or other computational approaches on synthetic.spectra. For attribution-only approaches, T=S.

5. F, exposures inferred by computational approaches on synthetic.spectra.

Folder structure for SynSigEval v0.2

Summary function will fit to the new 5-level folder structure:

First Level - top.level.dir: dataset folder (e.g. "S.0.1.Rsq.0.1", "syn.pancreas"). All spectra datasets under any top.level.dir have the same exposure.

Second Level - ground.truth.exposure.dir: spectra folder: (e.g. "sp.sp", "sa.sa.96"). All spectra datasets under any second.level.dir have the same signature and the same exposure counts.

Third Level - third.level.dir: It can be ("Attr") for storing results of packages which can only do exposure attribution of known signatures ("Attr"); it can also be ("ExtrAttr"), folder to store results of software packages which can do de-novo extraction and following attribution.

Fourth Level - tool.dir: The results of a software package (e.g. "SigProExtractor.results","SignatureEstimation.QP.results"). Under this level, tool.dir may contain multiple run.dir, each is a run of the software package using a specific number of seed.

Fifth level - run.dir: contains results from a run of the software package using a specific number of seed. (e.g. "seed.1")

Summarize results

1. Summarize results in fifth-level run.dir:

   Relevant functions are:

   • SummarizeSigProExtractor

   • SignatureAnalyzerSummarizeTopLevel

   • SignatureAnalyzerSummarizeSBS1SBS5

   • SummarizeSigOneExtrAttrSubdir

   • SummarizeSigOneAttrSubdir

   • SummarizeSigOnehelmsmanSubdir

   • SummarizeSigOneSigProSSSubdir

2. Summarize results of multiple runs by a computational approach on one spectra data set:

   SummarizeMultiRuns

3. Summarize results of multiple computational approaches on one spectra data set:

   SummarizeMultiToolsOneDataset

4. Summarize results of multiple computational approaches on multiple spectra data sets:

   SummarizeMultiToolsMultiDatasets

WuyangFF95/SynSigEval documentation built on Sept. 18, 2022, 11:41 a.m.