Function: MutSig

MutSig

Determine significantly mutated genes in a set of genetic variations using MutSig. The algorithm takes genetic variations mapped to patients as input, and computes gene significance based on sample- and gene-specific background mutation rates (BMRs). Gene-specific BMRs are estimated using gene features (covariates) that may affect mutation rate, such as gene expression. BMR is further estimated separately for different mutation categories such as C->T, A->G, and indels. MutSig requires silent (and optionally noncoding) variants in addition to nonsilent variants, as they are used for estimating BMR.

The output is a prioritized list of significant genes with the following columns:

Gene: HUGO symbol.

p: nominal p-value.

pFDR: FDR-corrected p-value (should be used for determining significance).

Mutations{Nonsilent,Silent,Noncoding}: number of mutations in the gene having an effect (nonsilent) or not (silent, noncoding). Noncoding mutations may be left out if the data set contains too few of them.

Coverage{Nonsilent,Silent,Noncoding}: base pairs covered for the gene in each category. One locus may contribute to coverage in multiple categories if one of the three possible point mutations is silent and the other two are nonsilent.

NumNeighbors: number of genes in the covariate space neighborhood used for estimating the BMR for this gene.

MutationsNeighbors: silent and noncoding mutations in the genes in the neighborhood.

CoverageNeighbors: base pairs covered in the genes in the neighborhood.

NumPatients: number of distinct patients having a nonsilent mutation.

MutationRate = MutationsNonsilent / CoverageNonsilent

BackgroundRate = MutationsNeighbors / CoverageNeighbors

RateRatio = MutationRate / BackgroundRate. As a heuristic, ratios over 5 are highlighted in the report.

This component may be called using a variety of inputs. Mandatory inputs include only patient identifiers, variant locations and variant alleles. In this case, other information is inferred using ANNOVAR and custom logic. It is also possible to provide all necessary information a priori.

Genes are normally specified using HUGO as per MutSig convention. MutSig requires approx. 10 GB memory. See below for installation instructions.

Version	1.0
Bundle	sequencing
Categories	VariationAnalysis
Authors	Kristian Ovaska (kristian.ovaska@helsinki.fi)
Issue tracker	View/Report issues
Requires	MutSigCV ; MATLAB Compiler Runtime ; ANNOVAR ; Scala
Source files	component.xml function.scala
Usage	Example with default values MutSig( variants = CSV, //coverage = CSV, //covariates = CSV, altAlleleColumn = "ALT", annovarBin = "", annovarDB = "", builtinCovariates = "*", chromColumn = "CHROM", effectColumn = "", geneColumn = "", label = "MutSig", matlab = "", mutsig = "", patientColumn = <string>, positionColumn = "POS", refAlleleColumn = "REF" )

Version

1.0

Bundle

sequencing

Categories

VariationAnalysis

Authors

Kristian Ovaska (kristian.ovaska@helsinki.fi)

Issue tracker

View/Report issues

Requires

MutSigCV ; MATLAB Compiler Runtime ; ANNOVAR ; Scala

Source files

component.xml function.scala

Usage

Example with default values

Inputs

Name	Type	Mandatory	Description
variants	CSV	Mandatory	Raw variants. Must contains columns for patient identifiers, chromosome, chromosomal position and variant (alternative) allele. May also contain columns for gene identifiers (HUGO), predicted effect of variation, and reference allele. Column names are configured using parameters.
coverage	CSV	Optional	If given, contains genomic intervals for those parts of the genome that were selectively sequenced in the experiment. For example, an exome sequencing experiment would include the regions that were captured. The following columns must be present: Chromosome, Start (1-based), End (inclusive). Note: annotating coverage using ANNOVAR is relatively slow.
covariates	CSV	Optional	Gene covariate table. The first column contains gene identifiers (HUGO) and the rest of the columns contain numeric attributes of the genes.

Name

Type

Mandatory

Description

variants

CSV

Mandatory

Raw variants. Must contains columns for patient identifiers, chromosome, chromosomal position and variant (alternative) allele. May also contain columns for gene identifiers (HUGO), predicted effect of variation, and reference allele. Column names are configured using parameters.

coverage

CSV

Optional

If given, contains genomic intervals for those parts of the genome that were selectively sequenced in the experiment. For example, an exome sequencing experiment would include the regions that were captured. The following columns must be present: Chromosome, Start (1-based), End (inclusive). Note: annotating coverage using ANNOVAR is relatively slow.

covariates

CSV

Optional

Gene covariate table. The first column contains gene identifiers (HUGO) and the rest of the columns contain numeric attributes of the genes.

Outputs

Name	Type	Description
genes	CSV	Significance of mutated genes (for all defined genes). Contains the columns: Gene (HUGO); p (nominal p-value); pFDR (FDR-corrected p-value), as well as MutSig metrics and copies of covariates.
genesExcel	Excel	The genes output as a formatted Excel.

Name

Type

Description

genes

CSV

Significance of mutated genes (for all defined genes). Contains the columns: Gene (HUGO); p (nominal p-value); pFDR (FDR-corrected p-value), as well as MutSig metrics and copies of covariates.

genesExcel

Excel

The genes output as a formatted Excel.

Parameters

Name	Type	Default	Description
altAlleleColumn	string	"ALT"	In variants, column name for the variant allele. Must always be defined.
annovarBin	string	""	Path to ANNOVAR binary installation directory. This directory contains convert2annovar.pl, table_annovar.pl, etc. Only needed when geneColumn or effectColumn are empty or custom coverage is provided. If empty, the environment variable ANNOVAR_HOME is used instead (when needed).
annovarDB	string	""	Path to ANNOVAR database directory. This directory often contains hgNN_X.{fa,idx,txt} files. If empty, the environment variable ANNOVAR_DB is used instead (when needed).
builtinCovariates	string	"*"	Comma-separated list of column names for builtin covariates located in gene.covariates.txt in the MutSig installation directory. These covariates are combined with custom covariates, if defined. The special value * selects all builtin covariates.
chromColumn	string	"CHROM"	In variants, column name for the chromosome. Must always be defined.
effectColumn	string	""	In variants, column name for variant effect. This is either the native MutSig effect (noncoding/nonsilent/silent/null); the Variant_Classification column in the MAF format; ANNOVAR output; or the AminoChange column in RikuRator export files. Non-MutSig effect specifications are converted to the native format. If empty, this is inferred using ANNOVAR
geneColumn	string	""	In variants, column name for gene identifiers (HUGO). If empty, this is inferred using ANNOVAR.
label	string	"MutSig"	Label for the experiment that is used as sheet name in the Excel report.
matlab	string	""	Path to the MATLAB compiler runtime directory. This directory contains the subdirectories bin, mcr, resources, etc. If empty, the environment variable MCRROOT is used instead.
mutsig	string	""	Path to MutSig installation directory. This directory contains the main MutSig MCR binaries (run_MutSigCV.sh, etc.) and supplementary data files available on the MutSig web site. Always needed is mutation_type_dictionary_file.txt. Depending on optional inputs, chr_files_hg19 (reference genome directory), exome_full192.coverage.txt and gene.covariates.txt may also be needed. If empty, the environment variable MUTSIG_HOME is used instead.
patientColumn	string	(no default)	In variants, column name for patient identifiers. Must always be defined.
positionColumn	string	"POS"	In variants, column name for chromosomal position. Must always be defined.
refAlleleColumn	string	"REF"	In variants, column name for the reference allele. If empty, this is inferred from mutsig/chr_files_hg19.

Name

Type

Default

Description

altAlleleColumn

string

"ALT"

In variants, column name for the variant allele. Must always be defined.

annovarBin

string

Path to ANNOVAR binary installation directory. This directory contains convert2annovar.pl, table_annovar.pl, etc. Only needed when geneColumn or effectColumn are empty or custom coverage is provided. If empty, the environment variable ANNOVAR_HOME is used instead (when needed).

annovarDB

string

Path to ANNOVAR database directory. This directory often contains hgNN_X.{fa,idx,txt} files. If empty, the environment variable ANNOVAR_DB is used instead (when needed).

builtinCovariates

string

"*"

Comma-separated list of column names for builtin covariates located in gene.covariates.txt in the MutSig installation directory. These covariates are combined with custom covariates, if defined. The special value * selects all builtin covariates.

chromColumn

string

"CHROM"

In variants, column name for the chromosome. Must always be defined.

effectColumn

string

In variants, column name for variant effect. This is either the native MutSig effect (noncoding/nonsilent/silent/null); the Variant_Classification column in the MAF format; ANNOVAR output; or the AminoChange column in RikuRator export files. Non-MutSig effect specifications are converted to the native format. If empty, this is inferred using ANNOVAR

geneColumn

string

In variants, column name for gene identifiers (HUGO). If empty, this is inferred using ANNOVAR.

label

string

"MutSig"

Label for the experiment that is used as sheet name in the Excel report.

matlab

string

Path to the MATLAB compiler runtime directory. This directory contains the subdirectories bin, mcr, resources, etc. If empty, the environment variable MCRROOT is used instead.

mutsig

string

Path to MutSig installation directory. This directory contains the main MutSig MCR binaries (run_MutSigCV.sh, etc.) and supplementary data files available on the MutSig web site. Always needed is mutation_type_dictionary_file.txt. Depending on optional inputs, chr_files_hg19 (reference genome directory), exome_full192.coverage.txt and gene.covariates.txt may also be needed. If empty, the environment variable MUTSIG_HOME is used instead.

patientColumn

string

(no default)

In variants, column name for patient identifiers. Must always be defined.

positionColumn

string

"POS"

In variants, column name for chromosomal position. Must always be defined.

refAlleleColumn

string

"REF"

In variants, column name for the reference allele. If empty, this is inferred from mutsig/chr_files_hg19.

Test cases

Test case	Parameters▼	IN variants	IN coverage	IN covariates	OUT genes	OUT genesExcel
case1	properties	variants	coverage	(missing)	(missing)	(missing)
	geneColumn=gene, patientColumn=patient, effectColumn=effect, refAlleleColumn=
case2_minimal	properties	variants	(missing)	(missing)	(missing)	(missing)
	patientColumn=patient, refAlleleColumn=, chromColumn=Chromosome, positionColumn=Start, altAlleleColumn=AltAllele
case3_effect	properties	variants	coverage	(missing)	(missing)	(missing)
	geneColumn=gene, patientColumn=patient, effectColumn=effect, refAlleleColumn=
case4_covariates	properties	variants	coverage	covariates	(missing)	(missing)
	geneColumn=gene, patientColumn=patient, effectColumn=effect, builtinCovariates=expr,reptime, refAlleleColumn=

Test case

Parameters▼

IN
variants

IN
coverage

IN
covariates

OUT
genes

OUT
genesExcel

case1

properties

variants

coverage

(missing)