Manta User Guide

Table of Contents

• Introduction
• Method Overview
• Capabilities
• Input requirements
• Outputs
• Run configuration and execution
• Extended use cases

Introduction

Manta calls structural variants (SVs) and indels from mapped paired-end sequencing reads. It is optimized for analysis of germline variation in small sets of individuals and somatic variation in tumor/normal sample pairs. Manta discovers, assembles and scores large-scale SVs, medium-sized indels and large insertions within a single efficient workflow. The method is designed for rapid analysis on standard compute hardware: NA12878 at 50x genomic coverage is analyzed in less than 20 minutes on a 20 core server, and most WGS tumor/normal analyses can be completed within 2 hours. Manta combines paired and split-read evidence during SV discovery and scoring to improve accuracy, but does not require split-reads or successful breakpoint assemblies to report a variant in cases where there is strong evidence otherwise. It provides scoring models for germline variants in small sets of diploid samples and somatic variants in matched tumor/normal sample pairs. There is experimental support for analysis of unmatched tumor samples as well (see details below). Manta accepts input read mappings from BAM or CRAM files and reports all SV and indel inferences in VCF 4.1 format. For extended methods and benchmarking details, please see the Manta preprint.

Method Overview

Manta divides the SV and indel discovery process into two primary steps: (1) scanning the genome to find SV associated regions and (2) analysis, scoring and output of SVs found in such regions.

1. Build breakend association graph In this step the entire genome is scanned to discover evidence of possible SVs and large indels. This evidence is enumerated into a graph with edges connecting all regions of the genome which have a possible breakend association. Edges may connect two different regions of the genome to represent evidence of a long-range association, or an edge may connect a region to itself to capture a local indel/small SV association. Note that these associations are more general than a specific SV hypothesis, in that many breakend candidates may be found on one edge, although typically only one or two candidates are found per edge.

2. Analyze graph edges to find SVs The second step is to analyze individual graph edges or groups of highly connected edges to discover and score SVs associated with the edge(s). The substeps of this process include inference of SV candidates associated with the edge, attempted assembly of the SVs breakends, scoring/genotyping and filtration of the SV under various biological models (currently diploid germline and somatic), and finally, output to VCF.

Capabilities

Manta is capable of detecting all structural variant types which are identifiable in the absence of copy number analysis and large-scale de-novo assembly. Detectable types are enumerated further below.

For each structural variant and indel, Manta attempts to assemble the breakends to basepair resolution and report the left-shifted breakend coordinate (per the VCF 4.1 SV reporting guidelines), together with any breakend homology sequence and/or inserted sequence between the breakends. It is often the case that the assembly will fail to provide a confident explanation of the data -- in such cases the variant will be reported as IMPRECISE, and scored according to the paired-end read evidence only.

The sequencing reads provided as input to Manta are expected to be from a paired-end sequencing assay which results in an innie orientation between the two reads of each sequence fragment, each presenting a read from the outer edge of the fragment insert inward.

Manta is primarily tested for whole-genome and whole-exome (or other targeted enrichement) sequencing assays on DNA. For these assays the following applications are supported:

• Joint analysis of small sets of diploid individuals (where 'small' means family-scale -- roughly 10 or fewer samples)
• Subtractive analysis of a matched tumor/normal sample pair
• Analysis of an individual tumor sample

For the first use case above, note that there is no specific restriction against using Manta for the joint analysis of larger cohorts, but this has not been extensively tested so there may be stability or call quality issues.

Per the final use case above, tumor samples can be analyzed without a matched normal sample. In this case no scoring function is available, but the supporting evidence counts and many filters can still be usefully applied.

RNA-Seq analysis is still in development and not fully supported. It can be configured with the --rna flag. This will adjust filtration levels and take other RNA-specific filtration and intron handling steps (more details are provided further below).

Detected variant classes

Manta is able to detect all variation classes which can be explained as novel DNA adjacencies in the genome. Simple insertion/deletion events can be detected down to a configurable minimum size cutoff (defaulting to 8). All DNA adjacencies are classified into the following categories based on the breakend pattern:

• Deletions
• Insertions
  • Fully-assembled insertions
  • Partially-assembled (ie. inferred) insertions
• Inversions
• Tandem Duplications
• Interchromosomal Translocations

Known Limitations

Manta should not be able to detect the following variant types:

• Dispersed duplications
• Most expansion/contraction variants of a reference tandem repeat
• Small inversions
  • The limiting size is not tested, but in theory detection falls off below ~200bases. So-called micro-inversions might be detected indirectly as combined insertion/deletion variants.
• Fully-assembled large insertions
  • The maximum fully-assembled insertion size should correspond to approximately twice the read-pair fragment size, but note that power to fully assemble the insertion should fall off to impractical levels before this size
  • Note that manta does detect and report very large insertions when the breakend signature of such an event is found, even though the inserted sequence cannot be fully assembled.

More general repeat-based limitations exist for all variant types:

• Power to assemble variants to breakend resolution falls to zero as breakend repeat length approaches the read size.
• Power to detect any breakend falls to (nearly) zero as the breakend repeat length approaches the fragment size.

Note that while Manta classifies novel DNA-adjacencies, it does not infer the higher level constructs implied by the classification. For instance, a variant marked as a deletion by manta indicates an intrachromosomal translocation with a deletion-like breakend pattern, however there is no test of depth, b-allele frequency or intersecting adjacencies to directly infer the SV type.

Input requirements

The sequencing reads provided as input to Manta are expected to be from a paired-end sequencing assay with an innie orientation between the two reads of each DNA fragment, each presenting a read from the outer edge of the fragment insert inward.

Manta can tolerate non-paired reads in the input, so long as sufficient paired-end reads exist to estimate the paired fragment size distribution. Non-paired reads will still be used in discovery, assembly and split-read scoring if their alignments (or SA tag split alignments) support a large indel or SV, or mismatch/clipping suggests a possible breakend location.

Manta requires input sequencing reads to be mapped by an external tool and provided as input in either BAM or CRAM format. Each input file must be coordinate sorted and indexed to produce asamtools/htslib-style index in a file named to match the input BAM or CRAM file with an additional '.bai', '.crai' or '.csi' filename extension.

At configuration time, at least one BAM or CRAM file must be provided for the normal or tumor sample. A matched tumor-normal sample pair can be provided as well. If multiple input files are provided for the normal sample, each file will be treated as a separate sample as part of a joint diploid sample analysis.

The following limitations exist on the input BAM or CRAM files provided to Manta:

• Alignments cannot contain the = character in the SEQ field.
• Alignments cannot use the sequence match/mismatch (=/X) CIGAR notation
• RG (read group) tags in the alignment records are ignored -- each file will be treated as representing one sample.
• Alignments with basecall quality values greater than 70 are rejected (these are not supported on the assumption that this indicates an offset error)

Manta also requires a reference sequence in fasta format. This must be the same reference used for mapping the input alignment files. The reference must include a samtools/htslib-style index in a file named to match the input fasta with an additional '.fai' file extension.

Outputs

Structural Variant predictions

The primary Manta outputs are a set of VCF 4.1 files, found in ${MANTA_ANALYSIS_PATH}/results/variants. Currently there are 3 VCF files created for a germline analysis, and an additional somatic VCF is produced for a tumor/normal subtraction. These files are:

• diploidSV.vcf.gz
  • SVs and indels scored and genotyped under a diploid model for the set of samples in a joint diploid sample analysis or for the normal sample in a tumor/normal subtraction analysis. In the case of a tumor/normal subtraction, the scores in this file do not reflect any information from the tumor sample.
• somaticSV.vcf.gz
  • SVs and indels scored under a somatic variant model. This file will only be produced if a tumor sample alignment file is supplied during configuration
• candidateSV.vcf.gz
  • Unscored SV and indel candidates. Only a minimal amount of supporting evidence is required for an SV to be entered as a candidate in this file. An SV or indel must be a candidate to be considered for scoring, therefore an SV cannot appear in the other VCF outputs if it is not present in this file. Note that by default this file includes indels down to a very small size (>= 8 bases). These are intended to be passed on to a small variant caller without scoring by manta itself (by default manta scoring starts at size 51).
• candidateSmallIndels.vcf.gz
  • Subset of the candidateSV.vcf.gz file containing only simple insertion and deletion variants of size 50 or less. Passing this file to a small variant caller like strelka or starling (Isaac Variant Caller) will provide continuous coverage over all indel sizes when the small variant caller and manta outputs are evaluated together. Alternate small indel candidate sets can be parsed out of the candidateSV.vcf.gz file if this candidate set is not appropriate.

For tumor-only analysis, Manta will produce an additional VCF:

• tumorSV.vcf.gz
  • Unscored SV and indel candidates (same content as the candidateSV.vcf.gz above), but including additional details: (1) paired and split read supporting evidence counts for each allele (2) a subset of the filters from the scored tumor-normal model are applied to the single tumor case to improve precision.

Manta VCF reporting format

Manta VCF output follows the VCF 4.1 spec for describing structural variants. It uses standard field names whereever possible. All custom fields are described in the VCF header. The section below highlights some of the variant representation details and lists the primary VCF field values.

VCF Sample Names

Sample names printed into the VCF output are extracted from each input alignment file from the first read group ('@RG') record found in the header. Any spaces found in the name will be replaced with underscores. If no sample name is found a default SAMPLE1, SAMPLE2, etc.. label will be used instead.

Small indels

All variants are reported in the VCF using symbolic alleles unless they are classified as a small indel, in which case full sequences are provided for the VCF REF and ALT allele fields. A variant is classified as a small indel if all of these criteria are met:

• The variant can be entirely expressed as a combination of inserted and deleted sequence.
• The deletion or insertion length is not 1000 or greater.
• The variant breakends and/or the inserted sequence are not imprecise.

When VCF records are printed in the small indel format, they will also include the CIGAR INFO tag describing the combined insertion and deletion event.

Insertions with incomplete insert sequence assembly

Large insertions are reported in some cases even when the insert sequence cannot be fully assembled. In this case Manta reports the insertion using the <INS> symbolic allele and includes the special INFO fields LEFT_SVINSSEQ and RIGHT_SVINSSEQ to describe the assembled left and right ends of the insert sequence. The following is an example of such a record from the joint diploid analysis of NA12878, NA12891 and NA12892 mapped to hg19:

chr1    11830208        MantaINS:1577:0:0:0:3:0 T       <INS>   999     PASS    END=11830208;SVTYPE=INS;CIPOS=0,12;CIEND=0,12;HOMLEN=12;HOMSEQ=TAAATTTTTCTT;LEFT_SVINSSEQ=TAAATTTTTCTTTTTTCTTTTTTTTTTAAATTTATTTTTTTATTGATAATTCTTGGGTGTTTCTCACAGAGGGGGATTTGGCAGGGTCACGGGACAACAGTGGAGGGAAGGTCAGCAGACAAACAAGTGAACAAAGGTCTCTGGTTTTCCCAGGCAGAGGACCCTGCGGCCTTCCGCAGTGTTCGTGTCCCTGATTACCTGAGATTAGGGATTTGTGATGACTCCCAACGAGCATGCTGCCTTCAAGCATCTGTTCAACAAAGCACATCTTGCACTGCCCTTAATTCATTTAACCCCGAGTGGACACAGCACATGTTTCAAAGAG;RIGHT_SVINSSEQ=GGGGCAGAGGCGCTCCCCACATCTCAGATGATGGGCGGCCAGGCAGAGACGCTCCTCACTTCCTAGATGTGATGGCGGCTGGGAAGAGGCGCTCCTCACTTCCTAGATGGGACGGCGGCCGGGCGGAGACGCTCCTCACTTTCCAGACTGGGCAGCCAGGCAGAGGGGCTCCTCACATCCCAGACGATGGGCGGCCAGGCAGAGACACTCCCCACTTCCCAGACGGGGTGGCGGCCGGGCAGAGGCTGCAATCTCGGCACTTTGGGAGGCCAAGGCAGGCGGCTGCTCCTTGCCCTCGGGCCCCGCGGGGCCCGTCCGCTCCTCCAGCCGCTGCCTCC  GT:FT:GQ:PL:PR:SR       0/1:PASS:999:999,0,999:22,24:22,32      0/1:PASS:999:999,0,999:18,25:24,20    0/0:PASS:230:0,180,999:39,0:34,0

VCF INFO Fields

VCF FORMAT Fields

VCF FILTER Fields

What do the values in Manta's VCF ID field mean?

The VCF ID or 'identifer' field can be used for annotation, or in the case of BND ('breakend') records for translocations, the ID value is used to link breakend mates or partners.

An example Manta VCF ID is MantaINS:1577:0:0:0:3:0. The value provided in this field reflects the SV association graph edge(s) from which the SV or indel was discovered. The ID value provided by Manta is primarily intended for internal use by manta developers. The value is guaranteed to be unique within any VCF file produced by Manta, and these ID values are used to link associated breakend records using the standard VCF MATEID key. The structure of this ID may change in the future, it is safe to use the entire value as a unique key, but parsing this value may lead to incompatibilities with future updates.

The exact meaning of the ID field for the current Manta version is described in the following section of the Manta developer guide.

Converting Manta VCF to BEDPE format

It can sometimes be convenient to express structural variants in BEDPE format. For such applications we recommend the script vcfToBedpe available from:

https://github.com/ctsa/svtools

This repository is forked from @hall-lab with edits to support VCF 4.1 SV format and match Manta's portability contstaints.

Note that BEDPE format greatly reduces structural variant information compared to Manta's VCF output. In particular breakend orientation, breakend homology and insertion sequence are lost, in addition to the ability to define fields for locus and sample specific information. For this reason we only recommend BEDPE as a temporary intermediate output for applications which require it.

Statistics

Additional secondary output is provided in ${MANTA_ANALYSIS_PATH}/results/stats

• alignmentStatsSummary.txt
  • fragment length quantiles for each input alignment file
• svLocusGraphStats.tsv
  • statistics and runtime information pertaining to the SV locus graph
• svCandidateGenerationStats.tsv
  • statistics and runtime information pertaining to the SV candidate generation
• svCandidateGenerationStats.xml
  • xml data backing the svCandidateGenerationStats.tsv report

Run configuration and execution

Manta is run in a two step procedure: (1) configuration and (2) workflow execution. The configuration step is used to specify the input data and any options pertaining to the variant calling methods themselves. The execution step is used to specify any parameters pertaining to how manta is executed (such as the total number of cores or SGE nodes over which the jobs should be parallelized). The second execution step can also be interrupted and restarted without changing the final result of the workflow.

Configuration

The workflow is configured with the script: ${MANTA_INSTALL_PATH}/bin/configManta.py . Running this script with no arguments will display all standard configuration options to folder. Note that all input alignment (BAM or CRAM) files and reference sequence must contain the same chromosome names in the same order. In addition all input alignment files and reference sequences must be indexed with samtools (or a utility which creates equivilent index files). Manta's default settings assume a whole genome DNA-Seq analysis, but there are configuration options for exome/targeted sequencing analysis in addition to RNA-Seq.

Single Diploid Sample Analysis -- Example Configuration:

${MANTA_INSTALL_PATH}/bin/configManta.py \
--bam NA12878_S1.bam \
--referenceFasta hg19.fa \
--runDir ${MANTA_ANALYSIS_PATH}

Joint Diploid Sample Analysis -- Example Configuration:

${MANTA_INSTALL_PATH}/bin/configManta.py \
--bam NA12878_S1.cram \
--bam NA12891_S1.cram \
--bam NA12892_S1.cram \
--referenceFasta hg19.fa \
--runDir ${MANTA_ANALYSIS_PATH}

Tumor Normal Analysis -- Example Configuration:

${MANTA_INSTALL_PATH}/bin/configManta.py \
--normalBam HCC1187BL.cram \
--tumorBam HCC1187C.cram \
--referenceFasta hg19.fa \
--runDir ${MANTA_ANALYSIS_PATH}

Tumor-Only Analysis -- Example Configuration:

${MANTA_INSTALL_PATH}/bin/configManta.py \
--tumorBam HCC1187C.cram \
--referenceFasta hg19.fa \
--runDir ${MANTA_ANALYSIS_PATH}

On completion, the configuration script will create the workflow run script ${MANTA_ANALYSIS_PATH}/runWorkflow.py. This can be used to run the workflow in various parallel compute modes per the instructions in the [Execution] section below.

Advanced configuration options

There are two sources of advanced configuration options:

• Options listed in the file: ${MANTA_INSTALL_PATH}/bin/configManta.py.ini
  • These parameters are not expected to change frequently. Changing the file listed above will re-configure all manta runs for the installation. To change parameters for a single run, copy the configManta.py.ini file to another location, change the desired parameter values and supply the new file using the configuration script's --config FILE option.
• Advanced options listed in: ${MANTA_INSTALL_PATH}/bin/configManta.py --allHelp
  • These options are intended primarily for workflow development and debugging, but could be useful for runtime optimization in some specialized cases.

Execution

The configuration step creates a new workflow run script in the requested run directory:

{MANTA_ANALYSIS_PATH}/runWorkflow.py

This script is used to control parallel execution of Manta via the pyFlow task engine. It can be used to parallelize structural variant analysis via one of two modes:

1. Parallelized across multiple cores on a single node.
2. Parallelized across multiple nodes on an SGE cluster.

A running workflow can be interrupted at any time and resumed where it left off. If desired, the resumed analysis can use a different running mode or total core count.

For a full list of execution options, see:

{MANTA_ANALYSIS_PATH}/runWorkflow.py -h

Example execution on a single node:

${MANTA_ANALYSIS_PATH}/runWorkflow.py -m local -j 8

Example execution on an SGE cluster:

${MANTA_ANALYSIS_PATH}/runWorkflow.py -m sge -j 36

Advanced execution options

These options are useful for Manta development and debugging:

• Stderr logging can be disabled with --quiet argument. Note this log is replicated to ${MANTA_ANALYSIS_PATH}/workspace/pyflow.data/logs/pyflow_log.txt so there is no loss of log information.
• The --rescore option can be provided to force the workflow to re-execute candidates discovery and scoring, but not the initial graph generation steps.

Extended use cases

Exome/Targeted

Supplying the '--exome' flag at configuration time will provide appropriate settings for WES and other regional enrichment analyses. At present this flag disables all high depth filters, which are designed to exclude pericentromeric reference compressions in the WGS case but cannot be applied correctly to a targeted analysis.

For small targeted regions, it may also be helpful to consider the high sensitivity calling documentation below.

Unpaired tumor sample

Manta supports SV calling for tumor sample only. The tumor-only mode can be triggered by supplying a tumor sample alignment file but no alignments for the normal sample. The results are reported in tumorSV.vcf.gz. This file contains all SV candidates (similar to the candidateSV.vcf.gz file), but also includes paired and split read evidence for each allele and a subset of the filters used for the tumor-normal comparative analysis. Note that Manta does not yet provide a quality scoring model for unpaired tumor sample analysis.

For low allele frequency variants, it may also be helpful to consider the high sensitivity calling documentation below.

RNA-Seq

Supplying the '--rna' flag at configuration time will provide experimental settings for RNA-Seq Fusion calling. At present this flag disables all high depth filters which are designed to exclude pericentromeric reference compressions in the WGS case but cannot be applied correctly to RNA-Seq analysis. In addition many custom RNA read processing and alignment steps are invoked. This mode is designed to function as part of larger workflow with additional steps to reduce overall false positive rate which take place downstream from Manta's fusion calling step.

It may also be helpful to consider the high sensitivity calling documentation below for this mode.

High sensitivity calling

Manta is configured with a discovery sensitivity appropriate for general WGS applications. In targeted or other specialized contexts the candidate sensitivity can be increased. A recommended general high sensitivity mode can be obtained by changing the two values 'minEdgeObservations' and 'minCandidateSpanningCount' in the manta configuration file (see 'Advanced configuration options' above) to 2 observations per candidate (the default is 3):

# Remove all edges from the graph unless they're supported by this many 'observations'.
# Note that one supporting read pair or split read usually equals one observation, but
# evidence is sometimes downweighted.
minEdgeObservations = 2

# Run discovery and candidate reporting for all SVs/indels with at least this
# many spanning support observations
minCandidateSpanningCount = 2