Uploaded September 2026 | Updated September 2026, 18 minutes ago
If you work on microbiome or metagenomics research, species-level labels can leave a critical question unanswered: which strain is actually doing the work?
In this PRISM 2026 presentation, Rob Knight, PhD, explains why quantitative long-read sequencing can change microbiome analysis from genus- or species-level association to strain identity, absolute counts, and mechanism. The talk moves from drug metabolism and MWAS examples to long-read data in Alzheimer's projects, Larry Smarr's 15-year time series, and environmental DNA monitoring.
"Wrong reads can't tell you. Long reads can. It's time to switch."
*Key takeaways*
- Same-species labels can hide large gene differences. Knight cites E. coli strains differing by 40% of their genes, with many other bacteria differing by at least 20%.
- Gut bacteria can chemically modify two-thirds of oral drugs, making strain, drug, and enzyme identity critical for prediction.
- In one MWAS example, the same people required six-fold fewer samples than a GWAS for traits including waist circumference, creatinine, and HbA1c.
- Long-read data helped produce 10-fold better contigs than short-read data on the same Malawi samples.
- Absolute quantitation with spike-in plasmids can move microbiome analysis from percentage of reads to CFU per milliliter.
- Scaled UniFrac work moved an analysis from 3,100 years of wall-clock time to minutes, enabling questions that were previously out of reach.
*Featured speakers*
- Rob Knight, Wolfe Family Endowed Chair in Microbiome Research at Rady Children's; Director, Center for Microbiome Innovation; Professor, University of California, San Diego
*Chapters*
00:05 Galileo, resolution, and microbiome causality
01:18 Why microbial species are not like human genomes
02:28 Drug metabolism needs strain-level answers
03:44 GWAS vs. microbiome-wide association studies
04:21 The Prevotella identity problem
05:03 Why short-read metagenomics breaks the puzzle
06:13 Long reads in Alzheimer's microbiome samples
06:53 Why species labels can hide function
08:04 Strain identity and transmission evidence
08:58 Scaling UniFrac, Qiita, and DartUniFrac
10:26 From Galileo's telescope to microbiome spectroscopy
11:43 Malawi long-read metagenomics and Larry Smarr's time series
13:33 Greengenes2 and backward-compatible long-read analysis
13:58 Absolute quantitation with spike-in plasmids
15:10 Pre- and post-surgery strain dynamics
16:41 Microbial dark matter and CRISPR
17:31 Integrated microbiome tools at UCSD
19:05 eDNA, Minderoo, and precision conservation
19:51 Function, mechanism, and human deployment
20:19 From wrong reads to long reads
*Resources*
- Microbial genomics with PacBio: pacb.com/microbial-genomics
- Microbiome and metagenome sequencing with HiFi reads: pacb.com/products-and-services/applications/complex-populations
- HiFi sequencing technology: pacb.com/technology/hifi-sequencing
- How HiFi sequencing works: pacb.com/technology/hifi-sequencing/how-it-works
Where would strain-level identity change your microbiome analysis first?
*Comment below.*
*Subscribe for more PacBio genomics content:*
youtube.com/@PacificBiosciences?sub_confirmation=1
Learn more about PacBio at pacb.com
Legal & Trademarks: pacb.com/legal-and-trademarks
For Research Use Only. Not for use in diagnostic procedures.
#PacBio #HiFiSequencing #Metagenomics
If you work on microbiome or metagenomics research, species-level labels can leave a critical question unanswered: which strain is actually doing the work?
In this PRISM 2026 presentation, Rob Knight, PhD, explains why quantitative long-read sequencing can change microbiome analysis from genus- or species-level association to strain identity, absolute counts, and mechanism. The talk moves from drug metabolism and MWAS examples to long-read data in Alzheimer's projects, Larry Smarr's 15-year time series, and environmental DNA monitoring.
"Wrong reads can't tell you. Long reads can. It's time to switch."
*Key takeaways*
- Same-species labels can hide large gene differences. Knight cites E. coli strains differing by 40% of their genes, with many other bacteria differing by at least 20%.
- Gut bacteria can chemically modify two-thirds of oral drugs, making strain, drug, and enzyme identity critical for prediction.
- In one MWAS example, the same people required six-fold fewer samples than a GWAS for traits including waist circumference, creatinine, and HbA1c.
- Long-read data helped produce 10-fold better contigs than short-read data on the same Malawi samples.
- Absolute quantitation with spike-in plasmids can move microbiome analysis from percentage of reads to CFU per milliliter.
- Scaled UniFrac work moved an analysis from 3,100 years of wall-clock time to minutes, enabling questions that were previously out of reach.
*Featured speakers*
- Rob Knight, Wolfe Family Endowed Chair in Microbiome Research at Rady Children's; Director, Center for Microbiome Innovation; Professor, University of California, San Diego
*Chapters*
00:05 Galileo, resolution, and microbiome causality
01:18 Why microbial species are not like human genomes
02:28 Drug metabolism needs strain-level answers
03:44 GWAS vs. microbiome-wide association studies
04:21 The Prevotella identity problem
05:03 Why short-read metagenomics breaks the puzzle
06:13 Long reads in Alzheimer's microbiome samples
06:53 Why species labels can hide function
08:04 Strain identity and transmission evidence
08:58 Scaling UniFrac, Qiita, and DartUniFrac
10:26 From Galileo's telescope to microbiome spectroscopy
11:43 Malawi long-read metagenomics and Larry Smarr's time series
13:33 Greengenes2 and backward-compatible long-read analysis
13:58 Absolute quantitation with spike-in plasmids
15:10 Pre- and post-surgery strain dynamics
16:41 Microbial dark matter and CRISPR
17:31 Integrated microbiome tools at UCSD
19:05 eDNA, Minderoo, and precision conservation
19:51 Function, mechanism, and human deployment
20:19 From wrong reads to long reads
*Resources*
- Microbial genomics with PacBio: pacb.com/microbial-genomics
- Microbiome and metagenome sequencing with HiFi reads: pacb.com/products-and-services/applications/complex-populations
- HiFi sequencing technology: pacb.com/technology/hifi-sequencing
- How HiFi sequencing works: pacb.com/technology/hifi-sequencing/how-it-works
Where would strain-level identity change your microbiome analysis first?
*Comment below.*
*Subscribe for more PacBio genomics content:*
youtube.com/@PacificBiosciences?sub_confirmation=1
Learn more about PacBio at pacb.com
Legal & Trademarks: pacb.com/legal-and-trademarks
For Research Use Only. Not for use in diagnostic procedures.
#PacBio #HiFiSequencing #Metagenomics
