Uploaded April 2026 | Updated September 2026, 1 day ago
If you run GWAS or interpret clinical genomes, the variants you've been ranking with short reads are probably not the variants doing the work.
Michael Schatz (Johns Hopkins) presents phase 1 of the All of Us long-read working group: 1,027 self-identified Black or African American participants sequenced on PacBio HiFi at 8x coverage, then scaled to 10,000 short-read participants via k-mer genotyping and imputation for a phenome-wide association study across 492 conditions. The headline finding: in roughly 2 of every 3 SV-trait associations, the structural variant is more statistically significant than any nearby SNP, and most are missing entirely from short-read call sets.
*Key takeaways*
- HiFi long-read sequencing detects 2 to 3 times more structural variants per person (~25,000) than state-of-the-art short-read pipelines like GATK-SV and Parliament2.
- Across 492 disease phenotypes, structural variants are statistically associated with most conditions; in about 2/3 of those associations, the SV beats the lead SNP.
- A linear callset of ~1.2M variants was paired with a strict 600,000-variant callset to power association testing without losing specificity.
- CADD-SV flagged ~300 highly pathogenic SVs in the cohort; ~170 overlap medically relevant genes, and most have never been seen before.
- Notable research findings include an SV truncating a cardiomyocyte-development gene associated with hypertensive heart failure and a stop-codon SV in a repetitive region associated with collapsed lung, both invisible to short-read GWAS.
- Phase 2 (10,000 HiFi-sequenced participants) is finishing now, with results coming at ASHG later this year.
- New SPRQ-Nx chemistry brings a HiFi human genome to ~$300–$350, a 30 to 40% reduction over current chemistry.
*Featured speakers*
- Michael Schatz, PhD — Bloomberg Distinguished Professor of Computer Science and Biology, Johns Hopkins University
- Aaron Wenger, PhD — Senior Director of Product Marketing, PacBio
- Nina Gonzaludo, PhD — Global Lead, Human Genomics Segment, PacBio (host)
*Chapters*
00:00 Webinar introduction and housekeeping
01:19 PacBio HiFi technology overview (Aaron Wenger)
04:46 Sawfish, SVX, and the CoLoRSdb frequency database
09:59 TRGT and the TRExplorer tandem-repeat catalog
11:46 SPRQ-Nx chemistry and the ~$300 HiFi genome
13:39 Introducing Michael Schatz
14:50 Why structural variants matter
17:59 SVs contribute more total variation than any other class
20:41 From single cases to population-scale SV studies
21:46 The All of Us research program
24:57 Phase 1: 1,027 HiFi-sequenced participants
29:48 Coverage, callsets, and variant calling at 8x
32:53 Population-scale variant detection vs short reads
34:24 KAGE k-mer genotyping and GLIMPSE imputation
39:48 CADD-SV pathogenicity scoring across the cohort
43:29 Cross-referencing the GWAS catalog
47:34 PheWAS across 492 disease phenotypes
50:50 Most disease variants are structural, not SNPs
54:14 Hypertensive heart failure: lead SV in a cardiomyocyte gene
55:27 Coding SVs and the collapsed-lung example
56:04 Summary: structural variants drive disease-associated traits
57:36 Phase 2 teaser and ASHG follow-up
58:11 Closing and Q&A handling
*Resources*
- PacBio Revio system: pacb.com/revio
- All of Us long-read study (PacBio blog): pacb.com/blog/landmark-all-of-us-study-demonstrates-hifi-sequencing-as-the-key-to-uncovering-disease-variants
- Population-scale long-read sequencing in All of Us (preprint, medRxiv): medrxiv.org/content/10.1101/2025.10.02.25336942v1
- CoLoRSdb (long-read SV frequency catalog): colorsdb.org
- Sawfish (joint SV/CNV caller): github.com/PacificBiosciences/sawfish
- SVX (cohort SV merging tool): github.com/PacificBiosciences/svx
If your cohort is still SNP-only, what's keeping long-read sequencing off the bench? *Comment below.*
*Subscribe for more PacBio research webinars and population-scale genomics talks.*
youtube.com/@PacificBiosciences?sub_confirmation=1
#LongReadSequencing #StructuralVariants #allofusaredead
Legal & Trademarks: pacb.com/legal-and-trademarks
For research use only. Not for use in diagnostic procedures.
If you run GWAS or interpret clinical genomes, the variants you've been ranking with short reads are probably not the variants doing the work.
Michael Schatz (Johns Hopkins) presents phase 1 of the All of Us long-read working group: 1,027 self-identified Black or African American participants sequenced on PacBio HiFi at 8x coverage, then scaled to 10,000 short-read participants via k-mer genotyping and imputation for a phenome-wide association study across 492 conditions. The headline finding: in roughly 2 of every 3 SV-trait associations, the structural variant is more statistically significant than any nearby SNP, and most are missing entirely from short-read call sets.
*Key takeaways*
- HiFi long-read sequencing detects 2 to 3 times more structural variants per person (~25,000) than state-of-the-art short-read pipelines like GATK-SV and Parliament2.
- Across 492 disease phenotypes, structural variants are statistically associated with most conditions; in about 2/3 of those associations, the SV beats the lead SNP.
- A linear callset of ~1.2M variants was paired with a strict 600,000-variant callset to power association testing without losing specificity.
- CADD-SV flagged ~300 highly pathogenic SVs in the cohort; ~170 overlap medically relevant genes, and most have never been seen before.
- Notable research findings include an SV truncating a cardiomyocyte-development gene associated with hypertensive heart failure and a stop-codon SV in a repetitive region associated with collapsed lung, both invisible to short-read GWAS.
- Phase 2 (10,000 HiFi-sequenced participants) is finishing now, with results coming at ASHG later this year.
- New SPRQ-Nx chemistry brings a HiFi human genome to ~$300–$350, a 30 to 40% reduction over current chemistry.
*Featured speakers*
- Michael Schatz, PhD — Bloomberg Distinguished Professor of Computer Science and Biology, Johns Hopkins University
- Aaron Wenger, PhD — Senior Director of Product Marketing, PacBio
- Nina Gonzaludo, PhD — Global Lead, Human Genomics Segment, PacBio (host)
*Chapters*
00:00 Webinar introduction and housekeeping
01:19 PacBio HiFi technology overview (Aaron Wenger)
04:46 Sawfish, SVX, and the CoLoRSdb frequency database
09:59 TRGT and the TRExplorer tandem-repeat catalog
11:46 SPRQ-Nx chemistry and the ~$300 HiFi genome
13:39 Introducing Michael Schatz
14:50 Why structural variants matter
17:59 SVs contribute more total variation than any other class
20:41 From single cases to population-scale SV studies
21:46 The All of Us research program
24:57 Phase 1: 1,027 HiFi-sequenced participants
29:48 Coverage, callsets, and variant calling at 8x
32:53 Population-scale variant detection vs short reads
34:24 KAGE k-mer genotyping and GLIMPSE imputation
39:48 CADD-SV pathogenicity scoring across the cohort
43:29 Cross-referencing the GWAS catalog
47:34 PheWAS across 492 disease phenotypes
50:50 Most disease variants are structural, not SNPs
54:14 Hypertensive heart failure: lead SV in a cardiomyocyte gene
55:27 Coding SVs and the collapsed-lung example
56:04 Summary: structural variants drive disease-associated traits
57:36 Phase 2 teaser and ASHG follow-up
58:11 Closing and Q&A handling
*Resources*
- PacBio Revio system: pacb.com/revio
- All of Us long-read study (PacBio blog): pacb.com/blog/landmark-all-of-us-study-demonstrates-hifi-sequencing-as-the-key-to-uncovering-disease-variants
- Population-scale long-read sequencing in All of Us (preprint, medRxiv): medrxiv.org/content/10.1101/2025.10.02.25336942v1
- CoLoRSdb (long-read SV frequency catalog): colorsdb.org
- Sawfish (joint SV/CNV caller): github.com/PacificBiosciences/sawfish
- SVX (cohort SV merging tool): github.com/PacificBiosciences/svx
If your cohort is still SNP-only, what's keeping long-read sequencing off the bench? *Comment below.*
*Subscribe for more PacBio research webinars and population-scale genomics talks.*
youtube.com/@PacificBiosciences?sub_confirmation=1
#LongReadSequencing #StructuralVariants #allofusaredead
Legal & Trademarks: pacb.com/legal-and-trademarks
For research use only. Not for use in diagnostic procedures.
