Mutect2 – GATK

Call somatic SNVs and indels via local assembly of haplotypes

Category Short Variant Discovery

Overview

Call somatic short mutations via local assembly of haplotypes. Short mutations include single nucleotide (SNA) and insertion and deletion (indel) alterations. The caller uses a Bayesian somatic genotyping model that differs from the original MuTect by Cibulskis et al., 2013 and uses the assembly-based machinery of HaplotypeCaller. Of note, Mutect2 v4.1.0.0 onwards enables joint analysis of multiple samples.

This tool is featured in the Somatic Short Mutation calling Best Practice Workflow. See Tutorial#11136 for a step-by-step description of the workflow and Article#11127 for an overview of what traditional somatic calling entails. For the latest pipeline scripts, see the Mutect2 WDL scripts directory. For pipelines with example data, see the gatk-workflows repository. Although we present the tool for somatic calling, it may apply to other contexts, such as mitochondrial variant calling and detection of somatic mosaicism.

Starting with v4.1.0.0 Mutect2 accomodates extreme high depths, e.g. 20,000X. See the following articles for details on this and additional applications.

Blog#23400 details general improvements to Mutect2 v4.1.0.0.
Blog#23598 details Mutect2 mitochondrial mode.
Blog#XXX (link to come) details use of Mutect2 in extremely low allele fraction variant detection.

Usage examples

Example commands show how to run Mutect2 for typical scenarios. The three modes are (i) tumor-normal mode where a tumor sample is matched with a normal sample in analysis, (ii) tumor-only mode where a single sample's alignment data undergoes analysis, and (iii) mitochondrial mode where sensitive calling at high depths is desirable.

As of v4.1, there is no longer a need to specify the tumor sample name with -tumor. You need only specify the normal sample name with -normal, if you include a normal.
Starting with v4.0.4.0, GATK recommends the default setting of --af-of-alleles-not-in-resource, which the tool dynamically adjusts for different modes. tumor-only calling sets the default to 5e-8, tumor-normal calling sets it to 1e-6 and mitochondrial mode sets it to 4e-3. For previous versions, the default was 0.001, the average heterozygosity of humans. For other organisms, change --af-of-alleles-not-in-resource to 1/(ploidy*samples in resource).

(i) Tumor with matched normal

Given a matched normal, Mutect2 is designed to call somatic variants only. The tool includes logic to skip emitting variants that are clearly present in the germline based on provided evidence, e.g. in the matched normal. This is done at an early stage to avoid spending computational resources on germline events. If the variant's germline status is borderline, then Mutect2 will emit the variant to the callset for subsequent filtering by FilterMutectCalls and review.

     gatk Mutect2 \
     -R reference.fa \
     -I tumor.bam \
     -I normal.bam \
     -normal normal_sample_name \
     --germline-resource af-only-gnomad.vcf.gz \
     --panel-of-normals pon.vcf.gz \
     -O somatic.vcf.gz

Mutect2 also generates a stats file names [output vcf].stats. That is, in the above example the stats file would be named somatic.vcf.gz.stats and would be in the same folder as somatic.vcf.gz. As of GATK 4.1.1 this file is a required input to FilterMutectCalls.

As of v4.1 Mutect2 supports joint calling of multiple tumor and normal samples from the same individual. The only difference is that -I and -normal must be specified for the extra samples.

     gatk Mutect2 \
     -R reference.fa \
     -I tumor1.bam \
     -I tumor2.bam \
     -I normal1.bam \
     -I normal2.bam \
     -normal normal1_sample_name \
     -normal normal2_sample_name \
     --germline-resource af-only-gnomad.vcf.gz \
     --panel-of-normals pon.vcf.gz \
     -O somatic.vcf.gz

(ii) Tumor-only mode

This mode runs on a single type of sample, e.g. the tumor or the normal. To create a PoN, call on each normal sample in this mode, then use CreateSomaticPanelOfNormals to generate the PoN.

  gatk Mutect2 \
   -R reference.fa \
   -I sample.bam \
   -O single_sample.vcf.gz

To call mutations on a tumor sample, call in this mode using a PoN and germline resource. After FilterMutectCalls filtering, consider additional filtering by functional significance with Funcotator.

  gatk Mutect2 \
  -R reference.fa \
  -I sample.bam \
  --germline-resource af-only-gnomad.vcf.gz \
  --panel-of-normals pon.vcf.gz \
  -O single_sample.vcf.gz

(iii) Mitochondrial mode

Mutect2 automatically sets parameters appropriately for calling on mitochondria with the --mitochondria flag. Specifically, the mode sets --initial-tumor-lod to 0, --tumor-lod-to-emit to 0, --af-of-alleles-not-in-resource to 4e-3, and the advanced parameter --pruning-lod-threshold to -4.

  gatk Mutect2 \
  -R reference.fa \
  -L chrM \
  --mitochondria-mode \
  -I mitochondria.bam \
  -O mitochondria.vcf.gz

The mode accepts only a single sample, which can be provided in multiple files.

(iv) Force-calling mode

This mode force-calls all alleles in force-call-alleles.vcf in addition to any other variants Mutect2 discovers.

  gatk Mutect2 \
   -R reference.fa \
   -I sample.bam \
   -alleles force-call-alleles.vcf
   -O single_sample.vcf.gz

If the sample is suspected to exhibit orientation bias artifacts (such as in the case of FFPE tumor samples) one should also collect F1R2 metrics by adding an --f1r2-tar-gz argument as shown below. This file contains information that can then be passed to LearnReadOrientationModel, which generate an artifact prior table for each tumor sample for FilterMutectCalls to use.

 gatk Mutect2 \
  -R reference.fa \
  -I sample.bam \
  --f1r2-tar-gz f1r2.tar.gz \
  -O single_sample.vcf.gz

Notes

Mutect2 does not require a germline resource nor a panel of normals (PoN) to run, although both are recommended. The tool prefilters sites for the matched normal and the PoN.

If a variant is absent from a given germline resource, then the value for --af-of-alleles-not-in-resource is used as an imputed allele frequency. Below is an excerpt of a known variants resource with population allele frequencies.

     #CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO
      1       10067   .       T       TAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCC      30.35   PASS    AC=3;AF=7.384E-5
      1       10108   .       CAACCCT C       46514.32        PASS    AC=6;AF=1.525E-4
      1       10109   .       AACCCTAACCCT    AAACCCT,*       89837.27        PASS    AC=48,5;AF=0.001223,1.273E-4
      1       10114   .       TAACCCTAACCCTAACCCTAACCCTAACCCTAACCCCTAACCCTAACCCTAACCCTAACCCTAACCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCCTAACCCTAACCCTAAACCCTA  *,CAACCCTAACCCTAACCCTAACCCTAACCCTAACCCCTAACCCTAACCCTAACCCTAACCCTAACCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCTAACCCCTAACCCTAACCCTAAACCCTA,T      36728.97        PASS    AC=55,9,1;AF=0.001373,2.246E-4,2.496E-5
      1       10119   .       CT      C,*     251.23  PASS    AC=5,1;AF=1.249E-4,2.498E-5
      1       10120   .       TA      CA,*    14928.74        PASS    AC=10,6;AF=2.5E-4,1.5E-4
      1       10128   .       ACCCTAACCCTAACCCTAAC    A,*     285.71  PASS    AC=3,1;AF=7.58E-5,2.527E-5
      1       10131   .       CT      C,*     378.93  PASS    AC=7,5;AF=1.765E-4,1.261E-4
      1       10132   .       TAACCC  *,T     18025.11        PASS    AC=12,2;AF=3.03E-4,5.049E-5

Additional parameters that factor towards filtering are available in FilterMutectCalls. While the tool calculates normal-lod assuming a diploid genotype, it calculates normal-artifact-lod with the same approach it uses for tumor-lod, i.e. with a variable ploidy assumption.
- If the normal artifact log odds becomes large, then FilterMutectCalls applies the artifact-in-normal filter..
- The tumor log odds, which is calculated independently of any matched normal, determines whether to filter a tumor variant. Variants with tumor LODs exceeding the threshold pass filtering.

Additional Information

Read filters

These Read Filters are automatically applied to the data by the Engine before processing by Mutect2.

Mutect2 specific arguments

This table summarizes the command-line arguments that are specific to this tool. For more details on each argument, see the list further down below the table or click on an argument name to jump directly to that entry in the list.

Argument name(s)	Default value	Summary
Required Arguments
--input -I		BAM/SAM/CRAM file containing reads
--output -O		File to which variants should be written
--reference -R		Reference sequence file
Optional Tool Arguments
--af-of-alleles-not-in-resource -default-af	-1.0	Population allele fraction assigned to alleles not found in germline resource. Please see docs/mutect/mutect2.pdf fora derivation of the default value.
--alleles		The set of alleles to force-call regardless of evidence
--annotation -A		One or more specific annotations to add to variant calls
--annotation-group -G		One or more groups of annotations to apply to variant calls
--annotations-to-exclude -AX		One or more specific annotations to exclude from variant calls
--arguments_file		read one or more arguments files and add them to the command line
--assembly-region-out		Output the assembly region to this IGV formatted file
--assembly-region-padding	100	Number of additional bases of context to include around each assembly region
--base-quality-score-threshold	18	Base qualities below this threshold will be reduced to the minimum (6)
--callable-depth	10	Minimum depth to be considered callable for Mutect stats. Does not affect genotyping.
--cloud-index-prefetch-buffer -CIPB	-1	Size of the cloud-only prefetch buffer (in MB; 0 to disable). Defaults to cloudPrefetchBuffer if unset.
--cloud-prefetch-buffer -CPB	40	Size of the cloud-only prefetch buffer (in MB; 0 to disable).
--disable-bam-index-caching -DBIC	false	If true, don't cache bam indexes, this will reduce memory requirements but may harm performance if many intervals are specified. Caching is automatically disabled if there are no intervals specified.
--disable-sequence-dictionary-validation	false	If specified, do not check the sequence dictionaries from our inputs for compatibility. Use at your own risk!
--downsampling-stride -stride	1	Downsample a pool of reads starting within a range of one or more bases.
--f1r2-max-depth	200	sites with depth higher than this value will be grouped
--f1r2-median-mq	50	skip sites with median mapping quality below this value
--f1r2-min-bq	20	exclude bases below this quality from pileup
--f1r2-tar-gz		If specified, collect F1R2 counts and output files into this tar.gz file
--founder-id		Samples representing the population "founders"
--gcs-max-retries -gcs-retries	20	If the GCS bucket channel errors out, how many times it will attempt to re-initiate the connection
--gcs-project-for-requester-pays		Project to bill when accessing "requester pays" buckets. If unset, these buckets cannot be accessed. User must have storage.buckets.get permission on the bucket being accessed.
--genotype-germline-sites	false	(EXPERIMENTAL) Call all apparent germline site even though they will ultimately be filtered.
--genotype-pon-sites	false	Call sites in the PoN even though they will ultimately be filtered.
--germline-resource		Population vcf of germline sequencing containing allele fractions.
--graph-output -graph		Write debug assembly graph information to this file
--help -h	false	display the help message
--ignore-itr-artifacts	false	Turn off read transformer that clips artifacts associated with end repair insertions near inverted tandem repeats.
--initial-tumor-lod -init-lod	2.0	Log 10 odds threshold to consider pileup active.
--interval-merging-rule -imr	ALL	Interval merging rule for abutting intervals
--intervals -L		One or more genomic intervals over which to operate
--max-assembly-region-size	300	Maximum size of an assembly region
--max-population-af -max-af	0.01	Maximum population allele frequency in tumor-only mode.
--max-reads-per-alignment-start	50	Maximum number of reads to retain per alignment start position. Reads above this threshold will be downsampled. Set to 0 to disable.
--min-assembly-region-size	50	Minimum size of an assembly region
--min-base-quality-score -mbq	10	Minimum base quality required to consider a base for calling
--mitochondria-mode	false	Mitochondria mode sets emission and initial LODs to 0.
--native-pair-hmm-threads	4	How many threads should a native pairHMM implementation use
--native-pair-hmm-use-double-precision	false	use double precision in the native pairHmm. This is slower but matches the java implementation better
--normal-lod	2.2	Log 10 odds threshold for calling normal variant non-germline.
--normal-sample -normal		BAM sample name of normal(s), if any. May be URL-encoded as output by GetSampleName with -encode argument.
--panel-of-normals -pon		VCF file of sites observed in normal.
--pcr-indel-qual	40	Phred-scaled PCR SNV qual for overlapping fragments
--pcr-snv-qual	40	Phred-scaled PCR SNV qual for overlapping fragments
--pedigree -ped		Pedigree file for determining the population "founders"
--sites-only-vcf-output	false	If true, don't emit genotype fields when writing vcf file output.
--tumor-lod-to-emit -emit-lod	3.0	Log 10 odds threshold to emit variant to VCF.
--version	false	display the version number for this tool
Optional Common Arguments
--add-output-sam-program-record	true	If true, adds a PG tag to created SAM/BAM/CRAM files.
--add-output-vcf-command-line	true	If true, adds a command line header line to created VCF files.
--create-output-bam-index -OBI	true	If true, create a BAM/CRAM index when writing a coordinate-sorted BAM/CRAM file.
--create-output-bam-md5 -OBM	false	If true, create a MD5 digest for any BAM/SAM/CRAM file created
--create-output-variant-index -OVI	true	If true, create a VCF index when writing a coordinate-sorted VCF file.
--create-output-variant-md5 -OVM	false	If true, create a a MD5 digest any VCF file created.
--disable-read-filter -DF		Read filters to be disabled before analysis
--disable-tool-default-read-filters	false	Disable all tool default read filters (WARNING: many tools will not function correctly without their default read filters on)
--exclude-intervals -XL		One or more genomic intervals to exclude from processing
--gatk-config-file		A configuration file to use with the GATK.
--interval-exclusion-padding -ixp	0	Amount of padding (in bp) to add to each interval you are excluding.
--interval-padding -ip	0	Amount of padding (in bp) to add to each interval you are including.
--interval-set-rule -isr	UNION	Set merging approach to use for combining interval inputs
--lenient -LE	false	Lenient processing of VCF files
--QUIET	false	Whether to suppress job-summary info on System.err.
--read-filter -RF		Read filters to be applied before analysis
--read-index		Indices to use for the read inputs. If specified, an index must be provided for every read input and in the same order as the read inputs. If this argument is not specified, the path to the index for each input will be inferred automatically.
--read-validation-stringency -VS	SILENT	Validation stringency for all SAM/BAM/CRAM/SRA files read by this program. The default stringency value SILENT can improve performance when processing a BAM file in which variable-length data (read, qualities, tags) do not otherwise need to be decoded.
--seconds-between-progress-updates	10.0	Output traversal statistics every time this many seconds elapse
--sequence-dictionary		Use the given sequence dictionary as the master/canonical sequence dictionary. Must be a .dict file.
--tmp-dir		Temp directory to use.
--use-jdk-deflater -jdk-deflater	false	Whether to use the JdkDeflater (as opposed to IntelDeflater)
--use-jdk-inflater -jdk-inflater	false	Whether to use the JdkInflater (as opposed to IntelInflater)
--verbosity	INFO	Control verbosity of logging.
Advanced Arguments
--active-probability-threshold	0.002	Minimum probability for a locus to be considered active.
--adaptive-pruning-initial-error-rate	0.001	Initial base error rate estimate for adaptive pruning
--allele-informative-reads-overlap-margin	2	Likelihood and read-based annotations will only take into consideration reads that overlap the variant or any base no further than this distance expressed in base pairs
--allow-non-unique-kmers-in-ref	false	Allow graphs that have non-unique kmers in the reference
--bam-output -bamout		File to which assembled haplotypes should be written
--bam-writer-type	CALLED_HAPLOTYPES	Which haplotypes should be written to the BAM
--debug-assembly -debug	false	Print out verbose debug information about each assembly region
--disable-adaptive-pruning	false	Disable the adaptive algorithm for pruning paths in the graph
--disable-tool-default-annotations	false	Disable all tool default annotations
--dont-increase-kmer-sizes-for-cycles	false	Disable iterating over kmer sizes when graph cycles are detected
--dont-use-soft-clipped-bases	false	Do not analyze soft clipped bases in the reads
--emit-ref-confidence -ERC	NONE	Mode for emitting reference confidence scores (For Mutect2, this is a BETA feature)
--enable-all-annotations	false	Use all possible annotations (not for the faint of heart)
--force-active	false	If provided, all regions will be marked as active
--force-call-filtered-alleles -genotype-filtered-alleles	false	Force-call filtered alleles included in the resource specified by --alleles
--gvcf-lod-band -LODB	[-2.5, -2.0, -1.5, -1.0, -0.5, 0.0, 0.5, 1.0]	Exclusive upper bounds for reference confidence LOD bands (must be specified in increasing order)
--independent-mates	false	Allow paired reads to independently support different haplotypes. Useful for validations with ill-designed synthetic data.
--kmer-size	[10, 25]	Kmer size to use in the read threading assembler
--max-mnp-distance -mnp-dist	1	Two or more phased substitutions separated by this distance or less are merged into MNPs.
--max-num-haplotypes-in-population	128	Maximum number of haplotypes to consider for your population
--max-prob-propagation-distance	50	Upper limit on how many bases away probability mass can be moved around when calculating the boundaries between active and inactive assembly regions
--max-suspicious-reads-per-alignment-start	0	Maximum number of suspicious reads (mediocre mapping quality or too many substitutions) allowed in a downsampling stride. Set to 0 to disable.
--max-unpruned-variants	100	Maximum number of variants in graph the adaptive pruner will allow
--min-dangling-branch-length	4	Minimum length of a dangling branch to attempt recovery
--min-pruning	2	Minimum support to not prune paths in the graph
--minimum-allele-fraction -min-AF	0.0	Lower bound of variant allele fractions to consider when calculating variant LOD
--num-pruning-samples	1	Number of samples that must pass the minPruning threshold
--pair-hmm-gap-continuation-penalty	10	Flat gap continuation penalty for use in the Pair HMM
--pair-hmm-implementation -pairHMM	FASTEST_AVAILABLE	The PairHMM implementation to use for genotype likelihood calculations
--pcr-indel-model	CONSERVATIVE	The PCR indel model to use
--phred-scaled-global-read-mismapping-rate	45	The global assumed mismapping rate for reads
--pruning-lod-threshold	2.302585092994046	Ln likelihood ratio threshold for adaptive pruning algorithm
--recover-all-dangling-branches	false	Recover all dangling branches
--showHidden	false	display hidden arguments
--smith-waterman	JAVA	Which Smith-Waterman implementation to use, generally FASTEST_AVAILABLE is the right choice
Deprecated Arguments
--tumor-sample -tumor		BAM sample name of tumor. May be URL-encoded as output by GetSampleName with -encode argument.

Argument details

Arguments in this list are specific to this tool. Keep in mind that other arguments are available that are shared with other tools (e.g. command-line GATK arguments); see Inherited arguments above.

--active-probability-threshold

Minimum probability for a locus to be considered active.

double 0.002 [ [ -∞ ∞ ] ]

--adaptive-pruning-initial-error-rate

Initial base error rate estimate for adaptive pruning
Initial base error rate guess for the probabilistic adaptive pruning model. Results are not very sensitive to this parameter because it is only a starting point from which the algorithm discovers the true error rate.

double 0.001 [ [ -∞ ∞ ] ]

--add-output-sam-program-record / -add-output-sam-program-record

If true, adds a PG tag to created SAM/BAM/CRAM files.

boolean true

--add-output-vcf-command-line / -add-output-vcf-command-line

If true, adds a command line header line to created VCF files.

boolean true

--af-of-alleles-not-in-resource / -default-af

Population allele fraction assigned to alleles not found in germline resource. Please see docs/mutect/mutect2.pdf fora derivation of the default value.
Population allele fraction assigned to alleles not found in germline resource.

double -1.0 [ [ -∞ ∞ ] ]

--allele-informative-reads-overlap-margin

Likelihood and read-based annotations will only take into consideration reads that overlap the variant or any base no further than this distance expressed in base pairs

int 2 [ [ -∞ ∞ ] ]

--alleles

The set of alleles to force-call regardless of evidence

FeatureInput[VariantContext] null

--allow-non-unique-kmers-in-ref

Allow graphs that have non-unique kmers in the reference
By default, the program does not allow processing of reference sections that contain non-unique kmers. Disabling this check may cause problems in the assembly graph.

boolean false

--annotation / -A

One or more specific annotations to add to variant calls
Which annotations to include in variant calls in the output. These supplement annotations provided by annotation groups.

List[String] []

--annotation-group / -G

One or more groups of annotations to apply to variant calls
Which groups of annotations to add to the output variant calls. Any requirements that are not met (e.g. failing to provide a pedigree file for a pedigree-based annotation) may cause the run to fail.

List[String] []

--annotations-to-exclude / -AX

One or more specific annotations to exclude from variant calls
Which annotations to exclude from output in the variant calls. Note that this argument has higher priority than the -A or -G arguments, so these annotations will be excluded even if they are explicitly included with the other options.

List[String] []

--arguments_file

read one or more arguments files and add them to the command line

List[File] []

--assembly-region-out

Output the assembly region to this IGV formatted file
If provided, this walker will write out its assembly regions to this file in the IGV formatted TAB-delimited output: http://www.broadinstitute.org/software/igv/IGV Intended to make debugging the active region calculations easier

String null

--assembly-region-padding

Number of additional bases of context to include around each assembly region
Parameters that control assembly regions

int 100 [ [ -∞ ∞ ] ]

--bam-output / -bamout

File to which assembled haplotypes should be written
The assembled haplotypes and locally realigned reads will be written as BAM to this file if requested. Really for debugging purposes only. Note that the output here does not include uninformative reads so that not every input read is emitted to the bam. Turning on this mode may result in serious performance cost for the caller. It's really only appropriate to use in specific areas where you want to better understand why the caller is making specific calls. The reads are written out containing an "HC" tag (integer) that encodes which haplotype each read best matches according to the haplotype caller's likelihood calculation. The use of this tag is primarily intended to allow good coloring of reads in IGV. Simply go to "Color Alignments By > Tag" and enter "HC" to more easily see which reads go with these haplotype. Note that the haplotypes (called or all, depending on mode) are emitted as single reads covering the entire active region, coming from sample "HC" and a special read group called "ArtificialHaplotype". This will increase the pileup depth compared to what would be expected from the reads only, especially in complex regions. Note also that only reads that are actually informative about the haplotypes are emitted. By informative we mean that there's a meaningful difference in the likelihood of the read coming from one haplotype compared to its next best haplotype. If multiple BAMs are passed as input to the tool (as is common for M2), then they will be combined in the bamout output and tagged with the appropriate sample names. The best way to visualize the output of this mode is with IGV. Tell IGV to color the alignments by tag, and give it the "HC" tag, so you can see which reads support each haplotype. Finally, you can tell IGV to group by sample, which will separate the potential haplotypes from the reads. All of this can be seen in this screenshot

String null

--bam-writer-type

Which haplotypes should be written to the BAM
The type of BAM output we want to see. This determines whether HC will write out all of the haplotypes it considered (top 128 max) or just the ones that were selected as alleles and assigned to samples.

The --bam-writer-type argument is an enumerated type (WriterType), which can have one of the following values:

ALL_POSSIBLE_HAPLOTYPES: A mode that's for method developers. Writes out all of the possible haplotypes considered, as well as reads aligned to each
CALLED_HAPLOTYPES: A mode for users. Writes out the reads aligned only to the called haplotypes. Useful to understand why the caller is calling what it is

WriterType CALLED_HAPLOTYPES

--base-quality-score-threshold

Base qualities below this threshold will be reduced to the minimum (6)
Bases with a quality below this threshold will reduced to the minimum usable qualiy score (6).

byte 18 [ [ -∞ ∞ ] ]

--callable-depth

Minimum depth to be considered callable for Mutect stats. Does not affect genotyping.

int 10 [ [ -∞ ∞ ] ]

--cloud-index-prefetch-buffer / -CIPB

Size of the cloud-only prefetch buffer (in MB; 0 to disable). Defaults to cloudPrefetchBuffer if unset.

int -1 [ [ -∞ ∞ ] ]

--cloud-prefetch-buffer / -CPB

Size of the cloud-only prefetch buffer (in MB; 0 to disable).

int 40 [ [ -∞ ∞ ] ]

--create-output-bam-index / -OBI

If true, create a BAM/CRAM index when writing a coordinate-sorted BAM/CRAM file.

boolean true

--create-output-bam-md5 / -OBM

If true, create a MD5 digest for any BAM/SAM/CRAM file created

boolean false

--create-output-variant-index / -OVI

If true, create a VCF index when writing a coordinate-sorted VCF file.

boolean true

--create-output-variant-md5 / -OVM

If true, create a a MD5 digest any VCF file created.

boolean false

--debug-assembly / -debug

Print out verbose debug information about each assembly region

boolean false

--disable-adaptive-pruning

Disable the adaptive algorithm for pruning paths in the graph
A single edge multiplicity cutoff for pruning doesn't work in samples with variable depths, for example exomes and RNA. This parameter disables the probabilistic algorithm for pruning the assembly graph that considers the likelihood that each chain in the graph comes from real variation, and instead uses a simple multiplicity cutoff.

boolean false

--disable-bam-index-caching / -DBIC

If true, don't cache bam indexes, this will reduce memory requirements but may harm performance if many intervals are specified. Caching is automatically disabled if there are no intervals specified.

boolean false

--disable-read-filter / -DF

Read filters to be disabled before analysis

List[String] []

--disable-sequence-dictionary-validation / -disable-sequence-dictionary-validation

If specified, do not check the sequence dictionaries from our inputs for compatibility. Use at your own risk!

boolean false

--disable-tool-default-annotations / -disable-tool-default-annotations

Disable all tool default annotations
Hook allowing for the user to remove default annotations from the tool

boolean false

--disable-tool-default-read-filters / -disable-tool-default-read-filters

Disable all tool default read filters (WARNING: many tools will not function correctly without their default read filters on)

boolean false

--dont-increase-kmer-sizes-for-cycles

Disable iterating over kmer sizes when graph cycles are detected
When graph cycles are detected, the normal behavior is to increase kmer sizes iteratively until the cycles are resolved. Disabling this behavior may cause the program to give up on assembling the ActiveRegion.

boolean false

--dont-use-soft-clipped-bases

Do not analyze soft clipped bases in the reads

boolean false

--downsampling-stride / -stride

Downsample a pool of reads starting within a range of one or more bases.
Downsample a pool of reads starting within a range of one or more bases.

int 1 [ [ -∞ ∞ ] ]

--emit-ref-confidence / -ERC

Mode for emitting reference confidence scores (For Mutect2, this is a BETA feature)
The reference confidence mode makes it possible to emit a per-bp or summarized confidence estimate for a site being strictly homozygous-reference. See https://software.broadinstitute.org/gatk/documentation/article.php?id=4017 for information about GVCFs. For Mutect2, this is a BETA feature that functions similarly to the HaplotypeCaller reference confidence/GVCF mode.

The --emit-ref-confidence argument is an enumerated type (ReferenceConfidenceMode), which can have one of the following values:

NONE: Regular calling without emitting reference confidence calls.
BP_RESOLUTION: Reference model emitted site by site.
GVCF: Reference model emitted with condensed non-variant blocks, i.e. the GVCF format.

ReferenceConfidenceMode NONE

--enable-all-annotations

Use all possible annotations (not for the faint of heart)
You can use the -AX argument in combination with this one to exclude specific annotations. Note that some annotations may not be actually applied if they are not applicable to the data provided or if they are unavailable to the tool (e.g. there are several annotations that are currently not hooked up to HaplotypeCaller). At present no error or warning message will be provided, the annotation will simply be skipped silently. You can check the output VCF header to see which annotations were activated and thus might be applied (although this does not guarantee that the annotation was applied to all records in the VCF, since some annotations have additional requirements, e.g. minimum number of samples or heterozygous sites only -- see the documentation for individual annotations' requirements).

boolean false

--exclude-intervals / -XL

One or more genomic intervals to exclude from processing
Use this argument to exclude certain parts of the genome from the analysis (like -L, but the opposite). This argument can be specified multiple times. You can use samtools-style intervals either explicitly on the command line (e.g. -XL 1 or -XL 1:100-200) or by loading in a file containing a list of intervals (e.g. -XL myFile.intervals).

List[String] []

--f1r2-max-depth

sites with depth higher than this value will be grouped

int 200 [ [ -∞ ∞ ] ]

--f1r2-median-mq

skip sites with median mapping quality below this value

int 50 [ [ -∞ ∞ ] ]

--f1r2-min-bq

exclude bases below this quality from pileup

int 20 [ [ -∞ ∞ ] ]

--f1r2-tar-gz

If specified, collect F1R2 counts and output files into this tar.gz file

File null

--force-active

If provided, all regions will be marked as active

boolean false

--force-call-filtered-alleles / -genotype-filtered-alleles

Force-call filtered alleles included in the resource specified by --alleles

boolean false

--founder-id / -founder-id

Samples representing the population "founders"

List[String] []

--gatk-config-file

A configuration file to use with the GATK.

String null

--gcs-max-retries / -gcs-retries

If the GCS bucket channel errors out, how many times it will attempt to re-initiate the connection

int 20 [ [ -∞ ∞ ] ]

--gcs-project-for-requester-pays

Project to bill when accessing "requester pays" buckets. If unset, these buckets cannot be accessed. User must have storage.buckets.get permission on the bucket being accessed.

String ""

--genotype-germline-sites

(EXPERIMENTAL) Call all apparent germline site even though they will ultimately be filtered.
Usually we exclude sites in the panel of normals from active region determination, which saves time. Setting this to true causes Mutect to produce a variant call at these sites. This call will still be filtered, but it shows up in the vcf.

boolean false

--genotype-pon-sites

Call sites in the PoN even though they will ultimately be filtered.
Usually we exclude sites in the panel of normals from active region determination, which saves time. Setting this to true causes Mutect to produce a variant call at these sites. This call will still be filtered, but it shows up in the vcf.

boolean false

--germline-resource

Population vcf of germline sequencing containing allele fractions.
A resource, such as gnomAD, containing population allele frequencies of common and rare variants.

FeatureInput[VariantContext] null

--graph-output / -graph

Write debug assembly graph information to this file
This argument is meant for debugging and is not immediately useful for normal analysis use.

String null

--gvcf-lod-band / -LODB

Exclusive upper bounds for reference confidence LOD bands (must be specified in increasing order)
When Mutect2 is run in reference confidence mode with banding compression enabled (-ERC GVCF), homozygous-reference sites are compressed into bands of similar tumor LOD (TLOD) that are emitted as a single VCF record. See the FAQ documentation for more details about the GVCF format.

This argument allows you to set the TLOD bands. Mutect2 expects a list of strictly increasing TLOD values that will act as exclusive upper bounds for the TLOD bands. To pass multiple values, you provide them one by one with the argument, as in `-LODB -3.0 -LODB -2.0 -LODB -1.0` (this would set the TLOD bands to be `[-Inf, -3.0), [-3.0, -2.0), [-2.0, -1.0), [-1.0, Inf]`, for example).

Note that, unlike the GQ used by HaplotypeCaller GVCFs, here the reference calls with the highest confidence are the most negative.

List[Double] [-2.5, -2.0, -1.5, -1.0, -0.5, 0.0, 0.5, 1.0]

--help / -h

display the help message

boolean false

--ignore-itr-artifacts

Turn off read transformer that clips artifacts associated with end repair insertions near inverted tandem repeats.
When opposite ends of a fragment are inverted tandem repeats of each other, the sequence past one end may be copied onto the other during library prep. By default, Mutect2 identifies and clips these artifacts, which are especially prevalent when DNA is damaged as in the case of FFPE samples and ancient DNA.

boolean false

--independent-mates

Allow paired reads to independently support different haplotypes. Useful for validations with ill-designed synthetic data.

boolean false

--initial-tumor-lod / -init-lod

Log 10 odds threshold to consider pileup active.
Only variants with estimated tumor LODs exceeding this threshold will be considered active.

double 2.0 [ [ -∞ ∞ ] ]

--input / -I

BAM/SAM/CRAM file containing reads

R List[GATKPath] []

--interval-exclusion-padding / -ixp

Amount of padding (in bp) to add to each interval you are excluding.
Use this to add padding to the intervals specified using -XL. For example, '-XL 1:100' with a padding value of 20 would turn into '-XL 1:80-120'. This is typically used to add padding around targets when analyzing exomes.

int 0 [ [ -∞ ∞ ] ]

--interval-merging-rule / -imr

Interval merging rule for abutting intervals
By default, the program merges abutting intervals (i.e. intervals that are directly side-by-side but do not actually overlap) into a single continuous interval. However you can change this behavior if you want them to be treated as separate intervals instead.

The --interval-merging-rule argument is an enumerated type (IntervalMergingRule), which can have one of the following values:

ALL
OVERLAPPING_ONLY

IntervalMergingRule ALL

--interval-padding / -ip

Amount of padding (in bp) to add to each interval you are including.
Use this to add padding to the intervals specified using -L. For example, '-L 1:100' with a padding value of 20 would turn into '-L 1:80-120'. This is typically used to add padding around targets when analyzing exomes.

int 0 [ [ -∞ ∞ ] ]

--interval-set-rule / -isr

Set merging approach to use for combining interval inputs
By default, the program will take the UNION of all intervals specified using -L and/or -XL. However, you can change this setting for -L, for example if you want to take the INTERSECTION of the sets instead. E.g. to perform the analysis only on chromosome 1 exomes, you could specify -L exomes.intervals -L 1 --interval-set-rule INTERSECTION. However, it is not possible to modify the merging approach for intervals passed using -XL (they will always be merged using UNION). Note that if you specify both -L and -XL, the -XL interval set will be subtracted from the -L interval set.

The --interval-set-rule argument is an enumerated type (IntervalSetRule), which can have one of the following values:

UNION: Take the union of all intervals
INTERSECTION: Take the intersection of intervals (the subset that overlaps all intervals specified)

IntervalSetRule UNION

--intervals / -L

One or more genomic intervals over which to operate

List[String] []

--kmer-size

Kmer size to use in the read threading assembler
Multiple kmer sizes can be specified, using e.g. `--kmer-size 10 --kmer-size 25`.

List[Integer] [10, 25]

--lenient / -LE

Lenient processing of VCF files

boolean false

--max-assembly-region-size

Maximum size of an assembly region

int 300 [ [ -∞ ∞ ] ]

--max-mnp-distance / -mnp-dist

Two or more phased substitutions separated by this distance or less are merged into MNPs.
Two or more phased substitutions separated by this distance or less are merged into MNPs.

int 1 [ [ -∞ ∞ ] ]

--max-num-haplotypes-in-population

Maximum number of haplotypes to consider for your population
The assembly graph can be quite complex, and could imply a very large number of possible haplotypes. Each haplotype considered requires N PairHMM evaluations if there are N reads across all samples. In order to control the run of the haplotype caller we only take maxNumHaplotypesInPopulation paths from the graph, in order of their weights, no matter how many paths are possible to generate from the graph. Putting this number too low will result in dropping true variation because paths that include the real variant are not even considered. You can consider increasing this number when calling organisms with high heterozygosity.

int 128 [ [ -∞ ∞ ] ]

--max-population-af / -max-af

Maximum population allele frequency in tumor-only mode.
In tumor-only mode, we discard variants with population allele frequencies greater than this threshold.

double 0.01 [ [ -∞ ∞ ] ]

--max-prob-propagation-distance

Upper limit on how many bases away probability mass can be moved around when calculating the boundaries between active and inactive assembly regions

int 50 [ [ -∞ ∞ ] ]

--max-reads-per-alignment-start

Maximum number of reads to retain per alignment start position. Reads above this threshold will be downsampled. Set to 0 to disable.
Other parameters

int 50 [ [ -∞ ∞ ] ]

--max-suspicious-reads-per-alignment-start

Maximum number of suspicious reads (mediocre mapping quality or too many substitutions) allowed in a downsampling stride. Set to 0 to disable.
Maximum number of suspicious reads (mediocre mapping quality or too many substitutions) allowed in a downsampling stride.

int 0 [ [ -∞ ∞ ] ]

--max-unpruned-variants

Maximum number of variants in graph the adaptive pruner will allow
The maximum number of variants in graph the adaptive pruner will allow

int 100 [ [ -∞ ∞ ] ]

--min-assembly-region-size

Minimum size of an assembly region
Parameters that control active regions

int 50 [ [ -∞ ∞ ] ]

--min-base-quality-score / -mbq

Minimum base quality required to consider a base for calling
Bases with a quality below this threshold will not be used for calling.

byte 10 [ [ -∞ ∞ ] ]

--min-dangling-branch-length

Minimum length of a dangling branch to attempt recovery
When constructing the assembly graph we are often left with "dangling" branches. The assembly engine attempts to rescue these branches by merging them back into the main graph. This argument describes the minimum length of a dangling branch needed for the engine to try to rescue it. A smaller number here will lead to higher sensitivity to real variation but also to a higher number of false positives.

int 4 [ [ -∞ ∞ ] ]

--min-pruning

Minimum support to not prune paths in the graph
Paths with fewer supporting kmers than the specified threshold will be pruned from the graph. Be aware that this argument can dramatically affect the results of variant calling and should only be used with great caution. Using a prune factor of 1 (or below) will prevent any pruning from the graph, which is generally not ideal; it can make the calling much slower and even less accurate (because it can prevent effective merging of "tails" in the graph). Higher values tend to make the calling much faster, but also lowers the sensitivity of the results (because it ultimately requires higher depth to produce calls).

int 2 [ [ -∞ ∞ ] ]

--minimum-allele-fraction / -min-AF

Lower bound of variant allele fractions to consider when calculating variant LOD

double 0.0 [ [ -∞ ∞ ] ]

--mitochondria-mode

Mitochondria mode sets emission and initial LODs to 0.
Mitochondria mode changes default values for --tumor-lod-to-emit and --initial-tumor-lod to 0 to increase sensitivity. --tumor-sample is also not explicitly required in mitochondria mode since a single sample bam is expected as input. Mitochondria mode is also required in FilterMutectCalls if used here.

Boolean false

--native-pair-hmm-threads

How many threads should a native pairHMM implementation use

int 4 [ [ -∞ ∞ ] ]

--native-pair-hmm-use-double-precision

use double precision in the native pairHmm. This is slower but matches the java implementation better

boolean false

--normal-lod

Log 10 odds threshold for calling normal variant non-germline.
This is a measure of the minimum evidence to support that a variant observed in the tumor is not also present in the normal. Applies to normal data in a tumor with matched normal analysis. The default has been tuned for diploid somatic analyses. It is unlikely such analyses will require changing the default value. Increasing the parameter may increase the sensitivity of somatic calling, but may also increase calling false positive, i.e. germline, variants.

double 2.2 [ [ -∞ ∞ ] ]

--normal-sample / -normal

BAM sample name of normal(s), if any. May be URL-encoded as output by GetSampleName with -encode argument.

List[String] []

--num-pruning-samples

Number of samples that must pass the minPruning threshold
If fewer samples than the specified number pass the minPruning threshold for a given path, that path will be eliminated from the graph.

int 1 [ [ -∞ ∞ ] ]

--output / -O

File to which variants should be written

R File null

--pair-hmm-gap-continuation-penalty

Flat gap continuation penalty for use in the Pair HMM

int 10 [ [ -∞ ∞ ] ]

--pair-hmm-implementation / -pairHMM

The PairHMM implementation to use for genotype likelihood calculations
The PairHMM implementation to use for genotype likelihood calculations. The various implementations balance a tradeoff of accuracy and runtime.

The --pair-hmm-implementation argument is an enumerated type (Implementation), which can have one of the following values:

EXACT
ORIGINAL
LOGLESS_CACHING
AVX_LOGLESS_CACHING
AVX_LOGLESS_CACHING_OMP
EXPERIMENTAL_FPGA_LOGLESS_CACHING
FASTEST_AVAILABLE

Implementation FASTEST_AVAILABLE

--panel-of-normals / -pon

VCF file of sites observed in normal.
A panel of normals can be a useful (optional) input to help filter out commonly seen sequencing noise that may appear as low allele-fraction somatic variants.

FeatureInput[VariantContext] null

--pcr-indel-model

The PCR indel model to use
When calculating the likelihood of variants, we can try to correct for PCR errors that cause indel artifacts. The correction is based on the reference context, and acts specifically around repetitive sequences that tend to cause PCR errors). The variant likelihoods are penalized in increasing scale as the context around a putative indel is more repetitive (e.g. long homopolymer). The correction can be disabling by specifying '-pcrModel NONE'; in that case the default base insertion/deletion qualities will be used (or taken from the read if generated through the BaseRecalibrator). VERY IMPORTANT: when using PCR-free sequencing data we definitely recommend setting this argument to NONE.

The --pcr-indel-model argument is an enumerated type (PCRErrorModel), which can have one of the following values:

NONE: no specialized PCR error model will be applied; if base insertion/deletion qualities are present they will be used
HOSTILE: a most aggressive model will be applied that sacrifices true positives in order to remove more false positives
AGGRESSIVE: a more aggressive model will be applied that sacrifices true positives in order to remove more false positives
CONSERVATIVE: a less aggressive model will be applied that tries to maintain a high true positive rate at the expense of allowing more false positives

PCRErrorModel CONSERVATIVE

--pcr-indel-qual

Phred-scaled PCR SNV qual for overlapping fragments

int 40 [ [ -∞ ∞ ] ]

--pcr-snv-qual

Phred-scaled PCR SNV qual for overlapping fragments

int 40 [ [ -∞ ∞ ] ]

--pedigree / -ped

Pedigree file for determining the population "founders"

File null

--phred-scaled-global-read-mismapping-rate

The global assumed mismapping rate for reads
The phredScaledGlobalReadMismappingRate reflects the average global mismapping rate of all reads, regardless of their mapping quality. This term effects the probability that a read originated from the reference haplotype, regardless of its edit distance from the reference, in that the read could have originated from the reference haplotype but from another location in the genome. Suppose a read has many mismatches from the reference, say like 5, but has a very high mapping quality of 60. Without this parameter, the read would contribute 5 * Q30 evidence in favor of its 5 mismatch haplotype compared to reference, potentially enough to make a call off that single read for all of these events. With this parameter set to Q30, though, the maximum evidence against any haplotype that this (and any) read could contribute is Q30. Set this term to any negative number to turn off the global mapping rate.

int 45 [ [ -∞ ∞ ] ]

--pruning-lod-threshold

Ln likelihood ratio threshold for adaptive pruning algorithm
Log-10 likelihood ratio threshold for adaptive pruning algorithm.

double 2.302585092994046 [ [ -∞ ∞ ] ]

--QUIET

Whether to suppress job-summary info on System.err.

Boolean false

--read-filter / -RF

Read filters to be applied before analysis

List[String] []

--read-index / -read-index

Indices to use for the read inputs. If specified, an index must be provided for every read input and in the same order as the read inputs. If this argument is not specified, the path to the index for each input will be inferred automatically.

List[GATKPath] []

--read-validation-stringency / -VS

Validation stringency for all SAM/BAM/CRAM/SRA files read by this program. The default stringency value SILENT can improve performance when processing a BAM file in which variable-length data (read, qualities, tags) do not otherwise need to be decoded.

The --read-validation-stringency argument is an enumerated type (ValidationStringency), which can have one of the following values:

STRICT
LENIENT
SILENT

ValidationStringency SILENT

--recover-all-dangling-branches

Recover all dangling branches
By default, the read threading assembler does not recover dangling branches that fork after splitting from the reference. This argument tells the assembly engine to recover all dangling branches.

boolean false

--reference / -R

Reference sequence file

R GATKPath null

--seconds-between-progress-updates / -seconds-between-progress-updates

Output traversal statistics every time this many seconds elapse

double 10.0 [ [ -∞ ∞ ] ]

--sequence-dictionary / -sequence-dictionary

Use the given sequence dictionary as the master/canonical sequence dictionary. Must be a .dict file.

String null

--showHidden / -showHidden

display hidden arguments

boolean false

--sites-only-vcf-output

If true, don't emit genotype fields when writing vcf file output.

boolean false

--smith-waterman

Which Smith-Waterman implementation to use, generally FASTEST_AVAILABLE is the right choice

The --smith-waterman argument is an enumerated type (Implementation), which can have one of the following values:

FASTEST_AVAILABLE: use the fastest available Smith-Waterman aligner that runs on your hardware
AVX_ENABLED: use the AVX enabled Smith-Waterman aligner
JAVA: use the pure java implementation of Smith-Waterman, works on all hardware

Implementation JAVA

--tmp-dir

Temp directory to use.

GATKPath null

--tumor-lod-to-emit / -emit-lod

Log 10 odds threshold to emit variant to VCF.
Only variants with tumor LODs exceeding this threshold will be written to the VCF, regardless of filter status. Set to less than or equal to tumor_lod. Increase argument value to reduce false positives in the callset. Default setting of 3 is permissive and will emit some amount of negative training data that {@link FilterMutectCalls} should then filter.

double 3.0 [ [ -∞ ∞ ] ]

--tumor-sample / -tumor

BAM sample name of tumor. May be URL-encoded as output by GetSampleName with -encode argument.

String null

--use-jdk-deflater / -jdk-deflater

Whether to use the JdkDeflater (as opposed to IntelDeflater)

boolean false

--use-jdk-inflater / -jdk-inflater

Whether to use the JdkInflater (as opposed to IntelInflater)

boolean false

--verbosity / -verbosity

Control verbosity of logging.

The --verbosity argument is an enumerated type (LogLevel), which can have one of the following values:

ERROR
WARNING
INFO
DEBUG

LogLevel INFO

--version

display the version number for this tool

boolean false

Return to top

GATK version 4.1.8.1-SNAPSHOT built at Sun, 2 Aug 2020 22:56:07 -0400.

2 comments

Leo

September 19, 2020 04:44
Hi, FYI in the first section, the link to "Blog#23400 details general improvements to Mutect2 v4.1.0.0" is broken.
0

Comment actions Permalink
Yen-Ping Yeh

February 19, 2021 07:23
Hello, I have a question about the tumor with matched normal. I used the mutect2 to call variant from single tumor and single normal (I had the three tumor samples, so I run three times). After filtering, annotation and merge. And I ran another way that is joint calling of multiple tumor and one normal sample. But I found that some variants in first way (Single tumor and single normal) could not be found in variants from joint calling.

Is joint calling stricter than single pair calling?

Sincerely,
Yen-Ping Yeh
0

Comment actions Permalink

Please sign in to leave a comment.

Genome Analysis Toolkit

Need Help?

Community Forum

Articles in this section

Category Short Variant Discovery

Overview

Usage examples

(i) Tumor with matched normal

(ii) Tumor-only mode

(iii) Mitochondrial mode

(iv) Force-calling mode

Notes

Additional Information

Read filters

Mutect2 specific arguments

Argument details

--add-output-sam-program-record / -add-output-sam-program-record

--add-output-vcf-command-line / -add-output-vcf-command-line

--af-of-alleles-not-in-resource / -default-af

--annotation / -A

--annotation-group / -G

--annotations-to-exclude / -AX

--bam-output / -bamout

--cloud-index-prefetch-buffer / -CIPB

--cloud-prefetch-buffer / -CPB

--create-output-bam-index / -OBI

--create-output-bam-md5 / -OBM

--create-output-variant-index / -OVI

--create-output-variant-md5 / -OVM

--debug-assembly / -debug

--disable-bam-index-caching / -DBIC

--disable-read-filter / -DF

--disable-sequence-dictionary-validation / -disable-sequence-dictionary-validation

--disable-tool-default-annotations / -disable-tool-default-annotations

--disable-tool-default-read-filters / -disable-tool-default-read-filters

--downsampling-stride / -stride

--emit-ref-confidence / -ERC

--exclude-intervals / -XL

--force-call-filtered-alleles / -genotype-filtered-alleles

--founder-id / -founder-id

--gcs-max-retries / -gcs-retries

--graph-output / -graph

--gvcf-lod-band / -LODB

--help / -h

--initial-tumor-lod / -init-lod

--input / -I

--interval-exclusion-padding / -ixp

--interval-merging-rule / -imr

--interval-padding / -ip

--interval-set-rule / -isr

--intervals / -L

--lenient / -LE

--max-mnp-distance / -mnp-dist

--max-population-af / -max-af

--min-base-quality-score / -mbq

--minimum-allele-fraction / -min-AF

--normal-sample / -normal

--output / -O

--pair-hmm-implementation / -pairHMM

--panel-of-normals / -pon

--pedigree / -ped

--read-filter / -RF

--read-index / -read-index

--read-validation-stringency / -VS

--reference / -R

--seconds-between-progress-updates / -seconds-between-progress-updates

--sequence-dictionary / -sequence-dictionary

--showHidden / -showHidden

--tumor-lod-to-emit / -emit-lod

--tumor-sample / -tumor

--use-jdk-deflater / -jdk-deflater

--use-jdk-inflater / -jdk-inflater

--verbosity / -verbosity

Related articles