Exploring germline variants that might play a role in tumor analysis

In somatic T/N analysis, DRAGEN uses a subtraction model for SNVs. VCF will only emit variants where there is no (or very limited) alternate reads in the normal sample; in other words, variants that are found in both the normal and somatic sample will NOT be emitted in the SNV/INDEL VCF. This means that most germline variants will be excluded from the somatic VCF. To extract germline variants, it is recommended to run an additional analysis, the DRAGEN Germline pipeline, on the normal sample. In BSSH, there is an option to Enable Small Variant Calling on Normal Samples, while running Somatic in Tumor-Normal mode. As the germline and somatic variants are output from separate analyses, the variants will need to be combined before further exploration or explored separately.

Will enabling germline tagging in T/N workflows provide germline variants in the SNV/INDEV VCF?

Only those somatic variants that match a known allele from a database. In T/N mode, where the VCF usually includes only somatic variants, then it is possible that some of these somatic variants may match a known allele from a database, and then be tagged as germline.

Germline variants can be labeled in the somatic VCF by adding a flag to the command line recipe or by checking to Enable Common Germline Variant Tagging in the BSSH/ICA applications. Note that this flag has implications for other settings. First, annotation must be enabled. Second, if TMB and germline tagging are both enabled, then the TMB database filter needs to be set to false. With these two flags, germline variants are tagged in the VCF but ignored during TMB calculation.

If I want both germline and somatic variants, should I run the Tumor sample in the Tumor/Only workflow, because that produces a VCF that contains both germline and somatic variants?

The matched normal provides a more reliable indicator of somatic variants than a database. Database germline tagging for SNV incorrectly tags ~2% of germline variants as somatic and falsely classifies some somatic variants as germline (i.e. false positives and false negatives for somatic SNVs). Additionally, there is limited support for differentiating between germline and somatic CNVs. If a paired normal sample is available, the highest performance for the detection of both somatic and germline variants is achieved by implementing the Tumor/Normal Somatic Variant Calling workflow (as described in this user guide) and also the Germline Caller on the Normal sample.

It is recommend to calculate MSI in tumor-only mode. Tumor-only mode relies on a reference directory of files (.dist) containing the microsatellite repeat distribution in a panel of normals. Both the reference directory and microsatellite list are provided - see the Resource Files page for more information - and need to match the reference genome used in the alignment step.

Distributed as DRAGEN secondary analysis Product Files

It is recommended that reference files are based on normal (non-tumorous) samples processed in the same methods used for clinical samples, matching library prep, sequencer instrumentation, etc. When that is not possible, product files to support the analysis of samples prepared with the Illumina FFPE DNA Prep with Exome 2.5 Enrichment are available mostly on the DRAGEN Resources page and detailed here. Generally, CNV and MSI resource files show good performance across sequencing instruments, while systematic noise files are highly dependent on the sequence instrument and therefore important to match with clinical samples.

Checking coverage on the target regions can be a quick way to determine if the map/align step ran as expected. Following the enrichment and sequencing guidelines, the Mean Target Coverage (MTC) Depth for tumor samples is expected to be ≥130x, for normal samples ≥50x. This metric is called "Average alignment coverage over target region" and can be found in the <sample>.target_bed_coverage_metrics_<tumor/normal>.csv file. Note that results for "Target bed" should be evaluated, as DRAGEN Report defaults to genome, while this is a target-based, WES, assay. Another metric to check is "Aligned reads in target region", which gives a percentage of the reads in the target region relative to the entire genome and can be thought as Percent Read Enrichment. Other quality metrics to check can include the percentage of Mapped Reads (aim of ≥ 90%) and the Insert length (aim of ≥ 125bp) in the <sample>.mapping_metrics.csv file.

llumina FFPE DNA Prep with Exome 2.5 Enrichment is part of an integrated whole-exome sequencing (WES) Tumor-Normal workflow to deliver variant calling and biomarker analysis in low-input formalin-fixed paraffin-embedded (FFPE) samples. This page provides software user guides to Illumina's cloud-based and on-premise solutions for the data analysis of this library preparation kit.

The software performs the following workflows to analyze sequencing data.

• DNA Mapping and Aligning

• Somatic Small Variant (SNV) Caller

• Copy Number Variant (CNV) Caller

• Tumor Mutational Burden (TMB)

• Microsatellite Instability (MSI)

• Homologous Recombination Deficiency (HRD)

• Variant annotation

The performance of the Illumina FFPE DNA Prep with Exome 2.5 Enrichment kit with tumor only samples is currently being evaluated. The DRAGEN v4.4 user guide provides a recipe for DNA Somatic Tumor-Only Solid WES UMI workflow. Settings specific for this kit are included here.

/opt/dragen/$VERSION/bin/dragen        #DRAGEN install path 
--ref-dir $REF_DIR                     #path to DRAGEN linear hashtable 
--output-directory $OUTPUT 
--intermediate-results-dir /staging    #tmp dir on fast SDD 
--output-file-prefix $PREFIX 
# Inputs
--tumor-fastq-list $PATH                #see 'Input Options' for FQ, BAM or CRAM 
--tumor-fastq-list-sample-id $STRING 
--fastq-list $PATH                      #see 'Input Options' for FQ, BAM or CRAM 
--fastq-list-sample-id $STRING 
# Mapper
--enable-map-align true
--enable-map-align-output true         #save the output BAM (default=false)
--qc-coverage-ignore-overlaps true     #do not double-count overlapping mates
--validate-pangenome-reference false   #currently the linear reference is recommended for somatic analysis
# UMI
--umi-enable true 
--umi-library-type nonrandom-duplex
--umi-min-supporting-reads 1
--umi-start-mask-length 1
--umi-end-mask-length 3
# Small variant caller 
--enable-variant-caller true 
--vc-target-bed $VC_TARGET_BED         #see Resource Files page
--vc-systematic-noise $PATH            #see Resource Files page
--vc-enable-umi-solid true 
--vc-enable-germline-tagging true 
--vc-enable-non-primary-allelic-filter true
--vc-enable-triallelic-filter false       
# CNV 
--enable-cnv true
--cnv-population-b-allele-vcf $POP_SNPs  #Path to population SNP VCF; see https://help.dragen.illumina.com/product-guide/dragen-v4.4/dragen-dna-pipeline/cnv-calling/additional-documentation/cnv-preprocessing#b-allele-counts-ascn-callers
--cnv-target-bed $VC_TARGET_BED 
--cnv-combined-counts $PATH            #see Resource Files page
# Annotation                           #annotation is required if enabling TMB
--enable-variant-annotation true
--variant-annotation-data $PATH        #see notes below
--variant-annotation-assembly GRCh37/8
# TMB 
--enable-tmb true
--tmb-enable-proxi-filter true
# HRD Scoring 
--enable-hrd true                       #requires CNV 
# Microsatellite Instability (MSI) 
--msi-command tumor-only
--max-base-quality 63                                    #default if UMI is enabled
--msi-coverage-threshold 40                              
--msi-microsatellites-file ${microsatellite_file}        #see Resource Files page
--msi-ref-normal-dir ${normal_reference_directory}       #see Resource Files page

DRAGEN Somatic App Version 4.4.6 in BaseSpace Sequence Hub is used to analyze sequencing data from llumina FFPE DNA Prep with Exome 2.5 Enrichment tumor and normal libraries. Refer to online support for general information about using the DRAGEN Somatic app. The settings defined here in this user guide are specific for this library preparation kit and for paired tumor-normal analysis (do not use for tumor only analysis).

Access to Resource Files

The analysis workflow uses resource files. See the Resource Files page for details on where to find those files, which should be manually uploaded to BSSH and available for analyses.

Analysis Settings

The following DRAGEN recipes are specific for Illumina FFPE DNA Prep with Exome 2.5 Enrichment tumor and normal libraries (do not use for tumor only analysis or for a different library prep method). A general recipe for Somatic analysis of Tumor Normal samples with UMI is provided on DRAGEN v4.4 support pages.

Notes and additional options

Hashtable

For DRAGEN somatic runs it is recommended to use the linear (non-graph) hashtable.

Please see:

Inputs

.

FQ list Input

or

FQ Input

or

BAM Input

or

CRAM Input

or

Mapper

UMI

The above recipe details UMI options specific for the Illumina FFPE DNA Prep with Exome 2.5 Enrichment Library kit.

SNV

The vc-sq-filter-threshold flag can be used to fine-tune the "somatic quality" value at which variants are called in order to balance sensitivity and specificity. Variants with values between the vc-sq-call-threshold and vc-sq-filter-threshold values are labeled as "weak_evidence" in the output VCF file. The for tumor-normal analysis is 17.5.

Annotation

Top directory containing Nirvana data file. Instructions on how to download the resource are at . DRAGEN is expecting a top directory containing the sub-directories Cache, References, Supplementary Annotation.

Microsatellite Instability (MSI)

Microsatellite Instability (MSI) is run in tumor-only mode, but other variants and biomarkers are calculated in tumor-normal mode. msi-coverage-threshold is a required parameter for both tumor-only and tumor-normal, and should be set according to the sample coverage. The sites in the msi-microsatellites-file need to match those in the dist files within the msi-ref-normal-dir .

Demo Data is available on BSSH with the Project name "NovaSeq6000/NovaSeqX: Illumina FFPE DNA Prep with Exome 2.5 Enrichment - Demo Data" and on ICA with the Bundle name "Illumina FFPE DNA Prep with Exome 2.5 Enrichment - Demo Data".

Data includes 14 samples, or seven Tumor/Normal pairs:

1. SNV-tumor-AF10-NovaSeq6K & SNV-normal-NovaSeq6K: Seracare Seraseq Tumor Mutation DNA Mix v2 AF10 and Seraseq WT (DNA/RNA) Reference Material. Samples with known truth variants at expected Variant Allele frequency at 10%, sequenced on the NovaSeq6000 Sequencing System.

2. SNV-tumor-AF5-NovaSeqX & SNV-normal-NovaSeqX: Seracare Seraseq Tumor Mutation DNA Mix v2 AF5 and Seraseq WT (DNA/RNA) Reference Material. Seracare Reference Material diluted to expected Variant Allele frequency at 5%, sequenced on the NovaSeqX Sequencing System.

3. CNV-tumor-NovaSeq6K & CNV-normal-NovaSeq6K: Seracare Seraseq Solid Tumor CNV Mix +3 Copies and Seraseq WT (DNA/RNA) Reference Material. Sample with known CNV events at expected duplication events with fold change at 3x, sequenced on the NovaSeq6000 Sequencing System.

4. FFPE-tumor-21706-NovaSeq6K & FFPE-normal-21707-NovaSeq6K: clinical formalin-fixed paraffin-embedded (FFPE) tumor sample 21706 and benign adjacent tissue sample 21707. FFPE samples sequenced on the NovaSeq6000 Sequencing System.

5. FFPE-tumor-21706-NovaSeqX & FFPE-normal-21707-NovaSeqX: clinical formalin-fixed paraffin-embedded (FFPE) tumor sample 21706 and benign adjacent tissue sample 21707. FFPE samples sequenced on the NovaSeqX Sequencing System.

6. FFPE-tumor-12293-exome-NovaSeq6K & FFPE-normal-12294-exome-NovaSeq6K: FFPE tumor sample 12293 and benign adjacent tissue sample 12294, enriched with the Twist Exome v2.5 panel. FFPE samples sequenced on the NovaSeq6000 Sequencing System.

7. FFPE-tumor-12293-exome-plus-spike-NovaSeq6K & FFPE-normal-12294-exome-plus-spike-NovaSeq6K: FFPE tumor sample 12293 and benign adjacent tissue sample 12294, enriched with the Twist Exome v2.5 panel and a custom spike-in panel. FFPE samples sequenced on the NovaSeq6000 Sequencing System.

Sample pair #7 (FFPE 12293/12294 enriched with and without a custom spike-in panel) was not included in variant calling because a systematic noise file and a Panel of Normals were not available for these combined covered regions. The pre-built WES systematic noise file based on the exome region (without spike-in) can be used, but precision in the custom spike-in areas may be reduced. The pre-built PON target counts based on the exome region (without spike-in) can be used, but CNV events may be inaccurate. Differences in coverage can seen by viewing the bam files.

The Illumina FFPE DNA Prep with Exome 2.5 Enrichment supports the addition of probes during the enrichment process. Spike-in probes are intended to increase coverage in Exome v2.5 regions or add coverage in areas of no coverage by the Exome 2.5 panel. The analysis workflow will be similar, but attention should be paid to the resource files.

Targeted Regions. This must be a BED file of the combined coverage areas, ie., Exome 2.5 Plus Panel and the custom panel. Refer to this.

Systematic noise file. If the spike-in panel adds coverage to new regions of the genome, it is recommended to generate a new systematic noise file from normal samples enriched with Exome v2.5 plus spike-in. Otherwise, variants calls in the new regions are more vulnerable to false positive calls than regions where, based on the systematic noise file, a somatic filter tackles noise that consistently appears at specific locations in the reference genome.

Panel of Normals (PON) for CNV calls. Whether the spike-in probes add coverage to new regions of the genome or increases coverage of coverage regions, it is always recommended to generate a custom PON, for your specific conditions, if calling CNVs. CNVs are called in regions characterized in the PON, so if the case sample has coverage in additional regions compared to the PON because of a spike-in, analysis will not call CNV events in those additional regions. However, the addition of spike-in probes can change enrichment broadly and may cause coverage fluctuations in previously characterized regions. Thus, CNV calls in regions that are shared by the PON and the case sample could be impacted by the addition of spike-in probes.

Use the DRAGEN_Somatic_Enrichment_4-3-17 Pipeline as part of the DRAGEN 4.3 Bundle in ICA to analyze sequencing data of llumina FFPE DNA Prep with Exome 2.5 Enrichment libraries. The settings defined here in this user guide are specific for this library preparation kit and should be used for paired tumor-normal samples.

Access to Resource Files

The analysis workflow uses resource files released with DRAGEN 4.4, so link both the DRAGEN 4.3 and DRAGEN 4.4 Bundles to your project. Additionally, Microsatellite Resource Files need to be manually uploaded and linked to your project if calculating MSI. See the Resource Files page for details on where to find those files. Note that the input for the Microsatellite Normal References can be a directory of individual distance files or combined distance file.

Analysis Settings

Start the DRAGEN Somatic Enrichment analysis

Under Pipelines, select DRAGEN_Somatic_Enrichment_4-3-17. Click Start Analysis on the top right of the page.

General

User Reference—A run name meaningful to the user

User tags, email notifications, and output folders are optional. If no output folder is selected, the output folder will be located in the root of the project.

Pricing

Select an Entitlement Bundle from the drop-down menu following Subscription.

Input files

Input the fastq (or ORA) files for the tumor and normal sample in a sample pair.

Reference

The reference genome to use for alignment. Reference genome files are located with the Illumina DRAGEN v10 Reference directory. The provided support the use of hg19-alt_masked.cnv.hla.methylation_combined.rna_v4.tar.gz and hg38-alt_masked.cnv.hla.methylation_combined.rna_v4.tar.gz.

Target BED file

BED file that contains targeted regions. If no spike-in probes were included in enrichment, select the Twist_ILMN_Exome_2.5_Panel bed file that matches the reference genome selected above, stored in the Illumina Enrichment BEDs directory.

If other spike-in probes were included in enrichment, generate a BED file of the combined coverage areas (ie., Exome 2.5 Plus Panel with the custom panel), upload to ICA, and select as the target BED file. See the Resources Files for information of format requirements.

Systematic Noise BED File

Select the systematic noise file that matches the panel used for enrichment and the reference genome used for alignment. The Illumina Systematic Noise directory within DRAGEN 4.4 Bundle contains WES_FFPE_NovaSeq_TwistV2.5_hg*_v1.0_systematic_noise.bed.gz. See the for information on pre-built and generating your own file.

CNV Panel of Normals/CNV Combined Counts

Select the files to comprise the target.counts baseline files. Files must be of the same type (eg, all .target.counts or all .target.counts.gc-corrected.). Select those files that match the panel used for enrichment, match the reference genome used for alignment, and has the desired handling of GC correction. The Illumina DRAGEN CNV Baseline Files/DRAGEN 4.4/ directory contains combined counts files for both hg19 and hg38. See the for information on pre-built target.count files.

MSI - Microsatellites File and Microsatellites Normal References Directory

As noted above, resource files for calculating MSI need to be uploaded to ICA and made available to the Project for inclusion in analysis.

Microsatellites File

Select the microsatellite list that maches the reference genome used for alignment: WES_v1.1.0_hg*_microsatellites.list

Microsatellites Normal References Directory/Combined Microsatellites Normal References File

Input the directory containing the individual distance files of the normal reference samples (hg*-WES-FFPE-*.microsat_normal.dist), or input a single merged file of the distances for each normal sample.

Settings

General Options

Output File Prefix—If something other than tumor is desired for the output file prefix, enter that here.

Sample Sex—Choose none from the drop-down list

Map Align Options

Enable Map/Align—true

Enable Map/Align Output—true

Enable Duplicate Marking—false

Map/Align Output—bam

Variant Calling Options

Enable Small Variant Caller—true

Leave Emit Ref Confidence and VCF File Output blank

Enable Germline Tagging—false

Enable CNV calling—true

CNV Use Somatic VC BAF—true

Enable SV calling—false

Targeted Callers

MSI command—tumor-only

MSI Coverage Threshold—60

If desired, set Enable Tumor Mutational Burden—true. Note that this requires variant annotation to be enabled.

Enable HLA—false

Enable HRD—false

Enable Tumor Mutational Burden—true

UMI Options

Enable UMI—true

UMI Library Type—nonrandom-duplex

UMI Aware Variant Calling—Low depth

Minimum Supporting UMI Reads—1

Variant Annotation Options

Enable Variant Annotation—true

Variant Annotation Assembly—reference corresponding to the genome used for variant calling

Additional Options

Additional DRAGEN Args—Paste the following commands into the box, making sure to update the name of the umi-metrics-interval-file

--umi-start-mask-length 1 --umi-end-mask-length 3 --qc-coverage-ignore-overlaps true --vc-sq-call-threshold 3 --vc-sq-filter-threshold 15

Resources

Select the resources setting

Start analysis

Once all parameters have been set, click Start analysis on the top right of the page

Rerun analysis settings with additional tumor-normal pairs

These settings can be rerun on additional tumor-normal sample pairs by selecting the particular analysis and clicking Rerun on the top right corner. Update the User reference and swap out the Tumor and Normal input files for the new pair.