Slide 1: DRAGEN v4.0: Unlock the Full Potential of Genomics
200026000_00_DRAGEN_v4.0_Customer_Webinar_Slides M-GL-01097 DRAGENTM v4.0: Unlock the Full Potential of Genomics Heidi Norton Sr. Product Manager, Illumina For Research Use Only. Not for use in diagnostic procedures.
Slide 2: DRAGEN for Secondary Analysis
Introduction DRAGEN for Secondary Analysis Library Prep Sequencing Secondary Analysis Tertiary Analysis and Reporting DRAGEN (Dynamic Read Analysis for GENomics) Accurate, comprehensive, and efficient secondary analysis For Research Use Only. Not for use in diagnostic procedures. 2
Slide 3: The right solution for your needs
The right solution for your needs Updated for DRAGEN v4.0 High Medium DRAGEN on-instrument DRAGEN on DRAGEN on Illumina DRAGEN Multi-Cloud DRAGEN on-premise (NextSeqTM 1000/2000) BaseSpaceTM Sequence Connected AnalyticsTM Bring Your Own License server Hub (BSSH) (ICA) (AWS, Azure) Low Push button so
Slide 4: DRAGEN v4.0 Recent Releases
Recent Releases DRAGEN v4.0 Unlock the full potential of genomics • Machine learning that improves small variant calling accuracy enabled by default More Accurate1 • Improved structural variant calling, HLA typing, mtDNA accuracy • Star Allele Caller identifies genotypes of 16 tier-1 pharmacogenes •
Slide 5: Unlock the full potential of genomics
Recent Releases Unlock the full potential of genomics DRAGEN v4.0 - Accurate, Comprehensive, Efficient Secondary Analysis Cell and Pharmacogenomics Genetic Disease Oncology Population Genomics Molecular Biology • New star allele • Expanded catalog of • New single-cell • HLA typing accuracy • 2x runt
Slide 6: DRAGEN improvements across applications
DRAGEN improvements across applications • Announcing Illumina DRAGEN Server v4 •• Machine Learning Announcing (ML) Illumina enabledServer DRAGEN by default v4 • Imputation pipeline (ML) Machine Learning (human reference more efficientonly) and enabled by default •• New imputation Structural Variantp
Slide 7: Announcing Illumina DRAGEN Server v4
Announcing Illumina DRAGEN Server v4 Latest version of DRAGEN server is here Purpose-built to provide the same great accurate, comprehensive, and efficient secondary analysis. • An incremental update to existing DRAGEN v3 server with moderately faster processing • Improved data center features • Com
Slide 8: Comparison between v3 and v4 servers
Comparison between v3 and v4 servers Illumina DRAGEN Server v3 Illumina DRAGEN Server v4 CPU Dual Intel Xeon Gold 6226 2.7 GHz, each with Dual Intel Xeon Gold 6226R 2.9 GHz, each with 12 cores. 24 cores total with 48 threads. 16 cores. 32 cores total with 64 threads. Memory 256 GB 512 GB Scratch Dri
Slide 9: Run time: just as fast, but more comprehensive
DRAGEN is just as fast, but more comprehensive DRAGEN run time, all callers 40 v3 server v4 server 35 35 34 • Illumina DRAGEN Server v4 is moderately 30 Run time, minutes 30 30 faster than v3 server 25 20 • DRAGEN v4.0 has similar run times to 15 DRAGEN v3.10 despite 4 additional callers 10 being en
Slide 10: Machine Learning (ML)
Machine Learning (ML) illumina
Slide 11: DRAGEN Machine Learning now enabled by default
More efficient DRAGEN Machine Learning (ML) now enabled by default Standard-Setting Accuracy Out of the Box ML introduced in v3.10 is now enabled by default with v4.0 for human germline small variant calling All Benchmark Regions Illumina DRAGEN v4.0 + Graph (ML by default) 99.9 Illumina DRAGEN v4.0
Slide 12: DRAGEN Machine Learning
DRAGEN Machine Learning ML introduced in v3.10 is now enabled by default with v4.0 for human germline small variant calling DRAGEN applies powerful machine learning Notes: techniques to refine human small variant calls and Non-ML DRAGEN-VC scores will no longer be emitted quality scores. as part of
Slide 13: DRAGEN-ML Accuracy
10 000 15 000 20 000 25 000 30 000 35 000 40 000 0 50 00 No -G ra ph 7472 15325 22797 No -G ra ph + ML 5866 15237 21103 Gr ap h 4735 4387 9122 HG 00 1 Gr ap h + M L 29713478 6449 No -G ra ph 8934 23176 32110 No -G ra ph + ML 6124 22411 28535 Gr ap h 5638 10086 15724 HG 00 2 Gr ap h + M L 2383 8146 1
Slide 14: DRAGEN Machine Learning pipeline
DRAGEN Machine Learning Integrated machine learning pipeline improves variant calling accuracy and is enabled by default Pretrained model + support files FASTQ Map/Align DRAGEN-VC DRAGEN-ML Recalibrated VCF Reference OR DRAGEN-ML will run automatically with VCF with updated QUAL, GT, BAM DRAGEN-VC G
Slide 15: Updates for Pharmacogenomics (PGx)
DRAGEN Updates for Pharmacogenomics (PGx) For Research Use Only. Not for use in diagnostic procedures.
Slide 16: Pharmacogenomics (PGx)
Pharmacogenomics (PGx) The study of how variations in the human genome dictate a person’s response to medications. In one study, more than 99% of people assessed had a genotype associated with a higher risk to at least one medication.1 1.Reisberg S et al. Translating genotype data of 44,000 biobank
Slide 17: Expand the value of your WGS analysis with DRAGEN PGx
Expand the value of your WGS analysis with DRAGEN PGx • Genotype difficult, but important PGx genes with CYP2D6 and CYP2B6 (new with v4.0) targeted callers • Determine optimal genotype and corresponding metabolism status associated with the genotype for “Tier 1” PGx genes (new with v4.0) For Researc
Slide 18: v4.0 PGx Overview
DRAGEN v4.0 PGx Overview DRAGEN PGx tools: 20 CPIC Level-A Genes 1. CACNA1S 2. CFTR New with v4.0 3. CYP2C19 1. Star allele caller, which calls star alleles 4. CYP2C9 5. CYP3A5 Improved with v4.0 and metabolism status for 16 genes 6. CYP4F2 Star alleles 7. IFNL3 and Star 8. RYR1 allele 9. NUDT15 met
Slide 19: Star Allele Caller
Star Allele Caller illumina
Slide 20: Star Allele Calling for PGx
Star Allele Calling for PGx Star alleles are haplotype patterns of a gene and can correlate with drug metabolism status for genes involved in drug metabolism 1.Reisberg S et al. Translating genotype data of 44,000 biobank participants into clinical pharmacogenetic recommendations: challenges and sol
Slide 21: DRAGEN Star Allele Caller
DRAGEN Star Allele Caller Call star alleles and metabolism status for 16 Tier 1 PGx Genes DRAGEN star allele caller can be enabled with an input The star allele caller identifies the genotypes of the gVCF file using the option –-enable-star-allele true. following 16 Tier-1 PGx genes: CACNA1S, CFTR,
Slide 22: Star Allele Caller pipeline
Star Allele Caller Star Allele Caller pipeline Genomic data (FASTQ/BAM/gVCF/VCF) Star allele Genotype genotyping PharmGKB Star allele Metabolism status definitions PharmCAT genotype- metabolism status "Gene": "CYP3A5”, mapping "Genotype": "*3/*3", "PharmCAT description": "An Primary input is a DRAGE
Slide 23: Star Allele Caller Validation
Star Allele Caller Validation Sample # Compare against Genes tested Concordance source samples DPYD, CACN1S, UGT1A1, TPMT, CYP3A5, 1K genomes PharmCAT 3201 CFTR, CYP2C19, CYP2C9, SLCO1B1, 100% project (from DRAGEN gVCF) NUDT15, VKORC1, CYP4F2, RYR1, IFNL3 PharmCAT DPYD, CACN1S, UGT1A1, TPMT, CYP3A5,
Slide 24: CYP2B6 Targeted Caller
CYP2B6 Targeted Caller illumina
Slide 25: CYP2B6 Targeted Caller
CYP2B6 Targeted Caller Star allele calling in important but difficult CYP2B6 gene The gene CYP2B6 is a member of the cytochrome DRAGEN CYP2B6 caller can be optionally enabled as P450 family of important pharmacogenes. It plays a part of the Germline pipeline using role in the metabolism of efavirenz
Slide 26: CYP2B6 Caller pipeline
CYP2B6 Caller CYP2B6 targeted calling pipeline Aligned Reads (FASTQ/BAM/CRAM) Total CYP2B6 and SV Calling to detect Small Variant Calling CYP2B7 copy number fusion genes Internal CYP2B6 caller configuration files Recombinant variant calling 18053A>G Requires WGS data with at least Gene conversion CY
Slide 27: CYP2B6 Caller Validation
CYP2B6 Caller Validation Sample source Star allele genotype Number of DRAGEN source Samples concordance Coriell samples GeT-RM1 76 70 (92.1%) 3 Coriell samples Stargazer2 70 70 (100%) 3 1K Genomes, PacBio HiFi reads + 125 123 (98.4%) 3,4 Illumina sequencing data manual inspection in IGV 1GeT-RM: htt
Slide 28: PGx Workflow
DRAGEN PGx Workflow Primary DRAGEN workflow DRAGEN PGx components External data DRAGEN DRAGEN FASTQ BAM gVCF read variant data data data mapping calling Star allele DRAGEN Targeted callers definitions from DRAGEN Star CYP2B6 | CYP2D6 | HLA PharmGKB & allele caller PharmVAR CYP2B6 CYP2D6 HLA Genotype
Slide 29: Updates for Genetic Disease
DRAGEN Updates for Genetic Disease For Research Use Only. Not for use in diagnostic procedures.
Slide 30: Repeat Expansions
Repeat Expansions Short Tandem Repeat (STR) expansions are a major cause of over 20 severe neurological disorders DRAGEN v4.0 with ExpansionHunter v5 expands the default catalog 2x and supports ~50k additional loci For Research Use Only. Not for use in diagnostic procedures. 30
Slide 31: Support for expanded catalog and runtime improvements
Support for expanded catalog and runtime improvements Runtime and memory usage improvements enabling scaling to large STR catalogs • Expansion Hunter v4 in DRAGEN 3.10 supported Example DRAGEN CMD line: genotyping of 30 STR loci --repeat-genotype-enable=true --repeat-genotype-use- catalog=<default|e
Slide 32: Expanded default catalog includes additional loci
Expanded default catalog includes 29 additional loci such as BEAN1, a challenging gene implicated in Spinocerebellar Ataxia AFF2 BEAN1 EIF4A3 JPH3 PRNP VWA1 AR C9ORF72 FMR1 LRP12 RAPGEF2 XYLT1 ARX_1 CACNA1A FOXL2 MARCHF6 RFC1 YEATS2 ARX_2 CBL FXN NIPA1 RUNX2 ZIC2 ATN1 CNBP GIPC1 NOP56 SAMD12 ZIC3 AT
Slide 33: Updates for Cell and Molecular Biology
DRAGEN Updates for Cell and Molecular Biology For Research Use Only. Not for use in diagnostic procedures.
Slide 34: Single-Cell ATAC-Seq
Single-Cell ATAC-Seq Enables single-cell resolution of chromatin accessibility + DRAGEN Single Cell ATAC can be run in combination with DRAGEN Single Cell RNA enabling simultaneous profiling of the transcriptome and epigenome For Research Use Only. Not for use in diagnostic procedures. 34
Slide 35: Single Cell Multiomics: RNA + ATAC
DRAGEN Single Cell Multiomics: RNA + ATAC DRAGEN Single Cell Multiomics pipeline enables simultaneous profiling of transcriptome and epigenome DRAGEN single-cell multiomics pipeline can be • Input: enabled using • ATAC input: FASTQ files with reads as triplets --enable-rna=true (ATAC read1, cell bar
Slide 36: Single-cell Multiomics: RNA + ATAC
Single-cell Multiomics: RNA + ATAC DRAGEN single-cell multiomics pipeline RNA Read alignment RNA Reads RNA Multiomics matrix (FASTQ) Read counting RNA barcode Error correction Multiomics Cell calling metrics DNA Read alignment ScATAC ATAC Reads ATAC fragments (FASTQ) Read counting ATAC barcode Error
Slide 37: Updates for Oncology
DRAGEN Updates for Oncology For Research Use Only. Not for use in diagnostic procedures.
Slide 38: Overview of Oncology updates
Overview of Oncology updates 1 2 3 4 HLA typing Fragmentomics Small variant Tumor only SV accuracy caller improvements improvements improvements Nucleotide-based HLA- Three new biomarkers • Improved efficiency enabling Accuracy improvements with specific alignments leverage quantifying genome-wide s
Slide 39: HLA genotyping
HLA genotyping illumina
Slide 40: HLA genotyping
HLA genotyping Genotyping HLA genes is notoriously difficult due to high levels of sequence homology between HLA genes and pseudogenes and dense variability DRAGEN v4.0 includes improved HLA typing accuracy For Research Use Only. Not for use in diagnostic procedures. 40
Slide 41: Overview of Class I HLA genotyping updates
Overview of Class I HLA genotyping updates 1 2 New HLA algorithm Accuracy boost across assay types • Nucleotide-based HLA- • WGS, WES > 99% accuracy specific alignments leverage DRAGEN map-align hardware • TSO500 panels > 97% accuracy • Expectation-maximization for genotyping *Note: HLA typing has n
Slide 42: New Class I HLA genotyping algorithm
New Class I HLA genotyping algorithm Nucleotide-based HLA-specific alignment and improved probabilistic modeling to infer HLA genotypes • Formerly, genotyping relied on amino-acid alignment Example DRAGEN CMD line: with maximal exact-match and integer linear programming dragen \ --enable-hla=true \
Slide 43: New and improved Class I HLA workflow
New and improved Class I HLA workflow Nucleotide-based alignments using DRAGEN map-align and expectation-maximization for genotyping Extract HLA reads: BED filter DRAGEN alignment against Process and filter alignments based on (auto-detected reference: IMGT/HLA Db: 7918 alleles DRAGEN alignment scor
Slide 44: High Class I HLA genotyping accuracy across sample types
High Class I HLA genotyping accuracy across sample types NGS assay DRAGEN mode Cohort Ground truth Overall accuracy Input type Reference genome 1000 Genomes Sanger sequencing Normal (germline) 99.6% BAM hg38_alt_masked_v2 n = 117 (exons 2 & 3) Internal cohort HISAT2 genotypes WGS Tumor/Normal pairs
Slide 45: Example command-lines
Example command-lines WGS/WES from BAM input: WGS/WES from FASTQ input: dragen \ dragen \ --enable-hla=true \ --enable-hla=true \ --output-directory={output_directory} \ --enable-map-align=true \ --output-file-prefix={prefix} \ --enable-sort=true \ --enable-map-align=true \ --output-directory={outpu
Slide 46: Example command-lines
Example command-lines WGS/WES Tumor-Normal mode from FASTQ input: dragen \ --enable-hla=true \ --enable-map-align=true \ --enable-sort=true \ --output-directory={output_directory} \ --output-file-prefix={prefix} \ --ref-dir={reference_directory} \ --tumor-fastq1={tumor_fq1} \ --tumor-fastq2={tumor_f
Slide 47: Example command-lines
Example command-lines TSO500-solid from FASTQ input: TSO500-solid from collapsed BAM input: dragen \ dragen \ --tso500-solid-hla=true \ --tso500-solid-hla=true \ --output-directory={output_directory} \ --output-directory={output_directory} \ --output-file-prefix={prefix} \ --output-file-prefix={pref
Slide 48: Example command-lines
Example command-lines TSO500-liquid from FASTQ input: TSO500-liquid from collapsed BAM input: dragen \ dragen \ --tso500-liquid-hla=true \ --tso500-liquid-hla=true \ --tso500-liquid=true \ --output-directory={output_directory} \ --output-directory={output_directory} \ --output-file-prefix={prefix} \
Slide 49: Fragmentomics
Fragmentomics illumina
Slide 50: Fragmentomics
Fragmentomics Fragmentomics is the study of fragmentation patterns of cell-free DNA or circulating tumor DNA. Fragmentation features are distinct between cancer and noncancer cells. It can be a powerful tool to infer cancer status and tissue of origin1. 1.Y. M. DENNIS LO, DIANA S. C. HAN, PEIYONG JI
Slide 51: DRAGEN Fragmentomics
DRAGEN Fragmentomics Fragmentomics metrics calculation for ctDNA assays Fragmentomics involves a set of biomarkers that are Example DRAGEN CMD line: useful for cancer detection and tissue-of-origin --enable-fragmentomics true --fragmentomics-exclude-bed hg19_exclude.bed --fragmentomics-bin-size 1000
Slide 52: DRAGEN Fragmentomics
DRAGEN Fragmentomics Fragmentomics metrics calculation for ctDNA assays End motif frequency Window protection store Fragment profile For Research Use Only. Not for use in diagnostic procedures.
Slide 53: DRAGEN Fragmentomics
DRAGEN Fragmentomics Fragmentomics metrics calculation for ctDNA assays End motif frequency Window protection store Fragment profile Get 5’ end motif of each read with motif Bed file containing 120bp windows Bin genome into user defined equal bins length specified by user specified by user bins Tabu
Slide 54: Fragmentomics Pipeline
DRAGEN Fragmentomics Pipeline Fragmentomics metrics calculation with ctDNA WGS tumor sample as input --fragmentomics-short-fragment-min-size --fragmentomics-short-fragment-max-size --fragmentomics-long-fragment-min-size --fragmentomics-long-fragment-max-size (Fragment size range) Tumor FASTQ MapAlig
Slide 55: DRAGEN Fragmentomics: >20x faster than existing pipelines
DRAGEN Fragmentomics > 20x faster than existing pipelines with high concordance WES ctDNA data downsampled to 10M. Fragment profile compared with Delfi. Motif WES ctDNA data downsampled to 10M. Fragment profile compared with Delfi. Motif frequency and window protection score compared with in house p
Slide 56: High Coverage Somatic Analyses
Small Variant Calling High Coverage Somatic Analyses More reliable variant calling on samples with high coverage As RAM is limited, DRAGEN can only hold a fixed Max supported amount of reads in memory at the same time. For Coverage* 3.10 4.0 somatic analyses, where high coverage can be beneficial to
Slide 57: Nucleotide (NTD) Error Bias Estimation (SNV)
Small Variant Calling Nucleotide (NTD) Error Bias Estimation (SNV) Improved error correction for FFPE • This feature corrects for sample-wide nucleotide SNV error in HCC1395 SEQQCII replicates substitution biases, e.g. increased C->T rate due to 700 oxidation/deamination that is common in FFPE • The
Slide 58: Methylation
Methylation illumina
Slide 59: Overview of methylation updates
Methylation Overview of methylation updates 1 2 End-to-end dedup UMI support • 1.5x run speed • Supports methylation assay with UMI barcode. • 1 run instead of 2 runs • Enables more accurate methylation calling by removing PCR error and sequencing error. For Research Use Only. Not for use in diagnos
Slide 60: Methylation: end-to-end dedup
Methylation: end-to-end dedup 1.5X faster and reduced overhead cost for dedup. Example DRAGEN CMD line: Previously, Getting sort/deduped cytosine report requires two runs: dragen --enable-map-align-output true --methylation-protocol directional Run1: Alignment: fastq->bam (deduped) --methylation-gen
Slide 61: UMI (Unique Molecular Identifiers) support
Methylation UMI (Unique Molecular Identifiers) support UMI enables more accurate methylation calling by removing PCR error and sequencing error • Methyl-UMI has the same principle and requirement as DNA UMI. Example DRAGEN CMD line: • Example: 5’ UMI-α UMI-β’ • Command line: P7 Index Primer Insert P
Slide 62: SV Tumor Only Scoring Model
DRAGEN SV Tumor Only Scoring Model Reduces Total Number of Reported SVs • DRAGEN SV T/O pipeline previously reported all COLO829 cell #PASS Events #PASS Events putative variants in the final VCF file, leading to low line SVs w/o scoring w/ scoring precision. model model • DRAGEN 4.0 introduces a T/O
Slide 63: SV Tumor Only Systematic Noise Filter
DRAGEN SV Tumor Only Systematic Noise Filter Improves Tumor Only Precision • Without a matched normal, the tumor only workflow COLO829 cell #PASS Events #PASS Events has challenges removing background noise from the line SVs w/o systematic w/ systematic analysis. noise filter noise filter • The DRAG
Slide 64: DRAGEN updates across applications
DRAGEN updates across applications
Slide 65: Imputation
Imputation illumina
Slide 66: Imputation
Imputation Imputation enables accurate genotyping of low-pass sequencing data, making WGS more accessible The GLIMPSE Imputation pipeline is integrated within DRAGEN v4.0 and https://odelaneau.github.io/GLIMPSE/overview.html is 2.3x faster For Research Use Only. Not for use in diagnostic procedures.
Slide 67: DRAGEN Imputation
Methylation DRAGEN Imputation New feature: end-to-end DRAGEN imputation tool Example DRAGEN CMD line: Integration of an software accelerated GLIMPSE 1 imputation tool into DRAGEN dragen --enable-imputation true --imputation-ref-panel-dir <REF_PANEL_DIR> Key Benefits: --imputation-chunk-input-region
Slide 68: End-to-end Imputation Pipeline
DRAGEN Imputation End-to-end Imputation Pipeline Imputation workflow BSSH Reference Panel chunk Reference Panel Partitions the genome into (IRPv1) (IRPv1) DRAGEN alignment and ”chunks” variant calling* Sample fastqs phase OR Performs imputation on a per-chunk basis List of VCFs* (.txt file) ligate L
Slide 69: DRAGEN Imputation improves sensitivity and reduces false positives
DRAGEN Imputation DRAGEN Imputation improves sensitivity of 1x data by >=70% and reduces false positives by >=80% 30x non-imputed 1x imputation 1x non-imputed - Imputed whole autosomal genome (~49M positions) with IRPv1 reference panel - All samples compared to NCBI vcf (GRCh38_1_22_v4.2.1, bi-allel
Slide 70: Multi-sample processing
DRAGEN Imputation Software accelerated code reduces computational time by ~2.3x Comparison of GLIMPSE vs DRAGEN BSSH end-to-end runtime: 50 samples*: 8 hours 44 minutes 36 seconds imputation pipeline runtimes 100 samples*: 15 hours 46 minutes 16 seconds Multi-sample processing 160 100 samples**: 21
Slide 71: DRAGEN multigenome reference builder
DRAGEN multigenome reference builder
Slide 72: Graph (Multigenome reference)
DRAGEN Graph (Multigenome reference) The DRAGEN graph genome (alt- aware multigenome hash table) contains population SNPs and alternate haplotypes to enable more accurate read mapping With DRAGEN v4.0, users can build their own multigenome reference For Research Use Only. Not for use in diagnostic p
Slide 73: Diversify your reference genome with multigenome reference builder
DRAGEN graph genome toolkit Diversify your reference genome with DRAGEN multigenome reference builder Build custom alt-aware multigenome hash tables for human genomes Example DRAGEN CMD line: Previously, DRAGEN graph (multigenome) hash tables were only available via download from the Illumina websit
Slide 74: DRAGEN multigenome reference builder
DRAGEN graph genome toolkit DRAGEN multigenome reference builder Build custom human multigenome hash tables Required input files: Phased VCFs pop alt contigs Graph BED1 Graph files alt contig liftover Hash table builder Custom graph generator hash table Reference FASTA 2 isolated pop SNPs Mask BED3
Slide 75: DRAGEN multigenome reference builder workflow
DRAGEN graph genome toolkit DRAGEN multigenome reference builder workflow Build custom human multigenome hash tables Generate isolated pop Generate pop alt contig SNPs and alt liftover Filter phased VCFs Filter phased VCFs Identify eligible SNPs within Identify eligible alt haplotypes graph BED with
Slide 76: ORA Compression
DRAGEN ORA Compression 9 8.25 8 DRAGEN ORA Compression 7 reduces FASTQ file sizes up to 5x 6 File size GB 5 3.87 4 3 2.61 With DRAGEN v4.0, users can 2 1.5 generate FASTQ.ORA files as 1 0.9 0.9 part of BCL Conversion 0 ENR RNA RNA SC-RNA scRNA Enrichment fastq.gz size fastq.ora size For Research Use
Slide 77: Generate compressed FASTQ.ORA files in BCL Conversion
Generate compressed FASTQ.ORA files in BCL Conversion BCL to FASTQ.ORA more user friendly • Convert BCL files directly to FASTQ.ORA (released Example DRAGEN CMD line: in v3.10 as a beta feature) dragen --bcl-conversion-only true • BCL conversion to FASTQ.ORA can now be --bcl-input-directory <DIR> --
Slide 78: FASTQ.ORA md5 checksum
FASTQ.ORA md5 checksum Compute md5 checksum of FASTQ.ORA during compression • Use case: to keep md5 checksum of FASTQ.ORA Example DRAGEN CMD line: file dragen --enable-map-align false • Enabled with option --ora-enable-md5 = true (false --ora-input <FILE> --enable-ora true by default) --ora-enable-m
Slide 79: Iterative gVCF Genotyper
Iterative gVCF Genotyper illumina
Slide 80: Iterative gVCF Genotyper updates in 4.0
DRAGEN PopGen Iterative gVCF Genotyper updates in 4.0 2x speed improvement, merge batch-wise msVCF, better HPC support Updates since DRAGEN 3.10 Example DRAGEN CMD line: • 2x speed improvement compared to DRAGEN 3.10 # aggregate a batch of gVCF files and write an msVCF # write concise msVCF output •
Slide 81: DRAGEN iterative gVCF Genotyper: speed improvements
DRAGEN PopGen DRAGEN iterative gVCF Genotyper: speed improvements 2x speed improvement compared to DRAGEN 3.10 mimalloc available at: github.com/microsoft/mimalloc Runtime [h] to process 200 germline gVCF using 48 threads DRAGEN release For Research Use Only. Not for use in diagnostic procedures. 81
Slide 82: DRAGEN compact gVCF
DRAGEN PopGen DRAGEN compact gVCF Compact gVCF germline input for gVCF Genotyper Variant entries have FORMAT/SPL, ICNT and GOP homref GT=0/0 variant default gVCF removed (only required for DRAGEN pedigree GQ bands: caller). 1 10 20 30 40 60 80 15 25 5 35 45 63 Hom-ref gVCF (genotype GT=0/0) entries
Slide 83: Merging of per-batch msVCF
DRAGEN PopGen Merging of per-batch msVCF New step to merge per-batch msVCF output batch msVCF 1 batch msVCF 2 batch msVCF 3 batch msVCF N …. Merge msVCF • If a combined msVCF of all batches is required, an additional step (--merge-batches) can be separately run to merge all the batch msVCF files int
Slide 84: Structural Variant caller improvements
Structural Variant caller improvements
Slide 85: Germline Structural Variant accuracy improvements
Germline Structural Variant accuracy improvements - Insertion recall improved by 2% compared to DRAGEN v3.10 - Recall now exceeds 72% and 64% for deletions and insertions respectively, while maintaining high precision. Method Type Recall Precision F-score Insertions 0.649 0.959 0.774 DRAGEN 4.0 Dele
Slide 86: Germline Structural Variant run time improvements
200026000_00_DRAGEN_v4.0_Customer_Webinar_Slides200026000 Germline Structural Variant run time improvements - Optimizations leading to improved analysis times across multiple SV workflows. Workflow Sample DRAGEN 3.10* DRAGEN 4.0* % Reduction Germline WGS NA12878 30x 5.5m 3m 45% Pedigree/Trio PlatGen
Slide 87: Mitochondrial variant calling
Mitochondrial variant calling
Slide 88: Mitochondrial variant calling
Mitochondrial variant calling Mitochondrial DNA Mitochondrial disorders are heterogeneous diseases that can arise at any stage of life. Studying mutations associated with mitochondrial disorders can be challenging due to phenotypic variability and genetic heterogeneity among individuals. DRAGEN v4.0
Slide 89: v4.0 improves mitochondrial variant calling accuracy
DRAGEN v4.0 improves mitochondrial variant calling accuracy Mitochondrial only DNA High allele • DRAGEN v4.0 improves frequency SNP variant calling SNP false negative accuracy in mitochondrial SNP false positive only DNA compared to DRAGEN v3.10 for both high and low allele frequencies Low allele fr
Slide 90: v4.0 improves mitochondrial variant calling accuracy
DRAGEN v4.0 improves mitochondrial variant calling accuracy SNP INDEL DRAGEN v4.0 improves SNP and INDEL variant calling accuracy in WGS data compared to DRAGEN v3.10 for both SNP and SNP FN INDEL FN INDEL SNP FP INDEL FP Platinum WGS data. Eberle et al. A reference data set of 5.4 million phased hu
Slide 91: DRAGEN BaseSpace Apps
DRAGEN BaseSpace Apps illumina
Slide 92: DRAGEN BaseSpace Apps
DRAGEN BaseSpace Apps • DRAGEN Imputation • Based on DRAGEN v3.7.8 • Based on DRAGEN v4.0.3 • Germline analysis for population genomics projects • Software accelerated GLIMPSE • VCF or FASTQ input • Built-in and custom reference panels • DRAGEN Single Cell ATAC • Runs gVCF Genotyper joint calling pi
Slide 93: DRAGEN BaseSpace Apps
200026000_00_DRAGEN_v4.0_Customer_Webinar_Slides DRAGEN BaseSpace Apps • DRAGEN Germline – All Callers • Based on DRAGEN v3.9.5 • All callers enabled • Integrated DRAGEN Reports • Accepts existing DRAGEN app results For Research Use Only. Not for use in diagnostic procedures. 93