This presentation introduces Roche's novel Sequencing by Expansion (SBX) Technology, a breakthrough approach designed for flexible, high-performance, and cost-efficient DNA sequencing. The technology merges Stratos Genomics' SBX chemistry, which converts DNA into an "expandimer" surrogate molecule for measurement, with Genia's High Throughput Sensor Array. Key innovations include custom expandable nucleotides (XNTPs), engineered polymerases (XP synthases), and a translocation control element for precise nanopore measurement.
The SBX platform demonstrates exceptional performance: duplex whole-genome sequencing (WGS) achieves Q39 accuracy, over 5 billion duplex reads in a one-hour run, and 30x coverage for seven samples in the same timeframe. An amplification-free "SBXFAST" protocol allows for a sample-to-VCF workflow in under 6.5 hours. The system supports a wide range of throughputs, from rapid runs to large-scale multi-omic experiments yielding 30 billion simplex reads in four hours, showcasing its versatility for both genomic and RNA applications. A reusable sensor module and accelerated local data processing contribute to its cost efficiency. Roche plans to release SBX as an open, research-use-only (RUO) product, targeting commercialization in 2026, with a vision for future clinical integration and continuous technological advancement.
Roche's Novel Sequencing by Expansion (SBX) Technology
0:00 Introduction to SBX: Roche introduces its novel Sequencing by Expansion Technology, highlighting its commitment to innovation and impact.
2:43 Future Sequencing Platform: The SBX platform is presented as an open, standalone sequencing solution for academic and translational research, with future plans for an end-to-end clinical setting.
3:22 SBX Inventor: Mark Kakoris, inventor of the SBX chemistry, is introduced, credited with developing the idea to address challenges in high-accuracy nanopore DNA sequencing.
5:15 Technology Design Principles: SBX is designed for flexibility, performance, and future scalability, focusing on accuracy, throughput, read length, and cost efficiency. It supports flexible operation (e.g., 20-minute to 4-hour runs).
6:17 Duplex WGS Accuracy: Demonstrates Q39 accuracy for duplex whole-genome sequencing (WGS) on reference sample HG001.
6:32 High Throughput: Achieves over 30x coverage for seven whole genomes in one hour, generating over 5 billion duplex reads.
7:00 Rapid Sample-to-VCF: Shows a whole-genome sequencing result from sample to VCF (variant call format) in less than seven hours.
7:12 Cost Efficiency: Emphasizes chemistry and measurement efficiencies, a reusable sensor module (sequencing chip), and efficient data analysis.
8:01 Technology Convergence: SBX chemistry from Stratos Genomics is combined with Genia's High Throughput Sensor Array Technology. The system uses separate instruments for chemistry synthesis and sequencing to maximize flexibility.
9:39 SBX Technology Overview: The approach does not directly sequence DNA but converts it into an "expandimer" – an expanded surrogate molecule – to rescale signal-to-noise challenges.
10:35 Foundational Preprint: A preprint detailing the SBX chemistry up to early 2020 is being made available.
11:08 SBX Chemistry Details: Involves expandable nucleotide triphosphates (XNTPs) with tether elements for selective cleavage and expansion, a translocation control element for reproducible movement through nanopores, and enhancer chemistries.
13:42 XP Synthase (Polymerase): An engineered polymerase, "XP or expandimer synthase," with an open architecture, showing over 10% amino acid mutations, 99.3% raw read simplex accuracy, processivity, and strand-displacing capabilities.
16:37 SBX Synthesis Workflow: A fixed solid-phase synthesis involves attaching an oligo, hybridizing the DNA library, synthesizing the expandimer, cleaving it, washing the chip, and then photo-cleaving to release the expandimer.
17:25 Translocation Control: A key innovation that allows for precise pausing and stepping of the expandimer through the nanopore, modulating throughput and yielding clear signal resolution.
19:17 8 Million Sensor Array: The technology scales from a single nanopore to an array of 8 million sensors, combining micro-wells, electrodes, and detection circuits. 7 to 7.5 million pores are typically functional.
20:47 Reusable Sensor Module: The sequencing sensor module is designed for significant reuse, contributing to overall efficiency.
22:15 Efficient Capture: A "leader concentrator" approach ensures efficient threading and sequencing of large expandimer molecules through the nanopore.
24:28 SBX Duplex Approach: Utilizes a hairpin Y-adapter structure to produce expandimers with both parent DNA strands, enabling intramolecular consensus for very high accuracy.
25:09 Duplex Workflow Details: Starts with 20-50 ng of unfragmented genomic DNA (or 2.5-10 ng cell-free DNA), uses linear amplification (not PCR) for improved accuracy, followed by expandimer synthesis and sequencing.
26:16 Duplex WGS Performance (7-plex, 1 hr): In a one-hour run, achieved 5.3 billion duplex reads (uniform yield), 34-38x coverage, and excellent F1 scores (SMV/Indel) with both Roche's and Google Deep Variant's callers.
27:39 Hartwig Medical Foundation Collaboration: Instruments were installed and qualified in two days in Hartwig's lab, demonstrating robust deployment. Preliminary data shows SBX SMV error rate 1.7x lower than Illumina.
31:19 SBXFAST Protocol: An amplification-free workflow designed for rapid turnaround, achieving library prep through VCF in ~5.5 hours.
32:11 SBXFAST Performance: Yields ~5 billion duplex reads in one hour with even better F1 scores, particularly for indels, and demonstrates strong performance across various variant types.
34:10 Broad Clinical Labs Collaboration: Similar rapid installation and validation at Broad, supporting the need for scalable, rapid clinical WGS.
37:07 Broad Fast Sequencing Demo: A recent demo (Feb 11th) achieved library prep through VCF for an HG1 sample in 6 hours and 25 minutes, with 20 minutes of sequencing, nearly 30x coverage, and respectable F1 scores.
38:43 Data Analysis Overview (John Mannion): Discusses SBXD read characteristics, including full-length and partial duplex reads, base quality definitions (concordant duplex, discordant duplex, simplex), and how these contribute to overall metrics (Q39 base accuracy, 30x coverage).
45:25 Insert Length Distribution: Shows a mean insert length of 235 base pairs with a significant long tail (300-400+ bp), valuable for mapping difficult genomic regions.
46:08 Accuracy Stability: Demonstrates stable accuracy across different insert lengths and positions within the read, with no significant degradation at read ends.
48:57 Homopolymer Accuracy: Achieves F1 scores greater than 99% for indel variants interacting with homopolymers up to 15 base pairs in length.
50:27 Coverage Uniformity: Exhibits good coverage uniformity across genomic regions, including challenging high GC content, a testament to chemistry optimization.
53:17 Accelerated Local Data Processing: Addressing the high data generation rates, the pipeline performs base calling in real-time and accelerates other steps (DMUX, consensus, mapping, variant calling) through online execution.
56:33 NVIDIA Parabricks Collaboration: Roche partners with NVIDIA to optimize Parabricks software components for the SBX pipeline.
57:46 Simplex Read Applications: Discusses applications for single-stranded reads, including 10x Flex probes and RNA isoform work, demonstrating impressive yields (e.g., ~14 billion CellRanger valid barcode reads per hour).
59:00 Simplex Yields: For 800bp HG1 templates, full-length reads average nearly 800bp with 99.36% accuracy, yielding 1.8 billion reads/hour and 1.4 terabases/hour.
1:02:43 Multi-Omic Experiment Output: In a four-hour run for a large-scale multi-omic experiment, 30 billion simplex reads were generated, showcasing SBX's capability for read-heavy applications.
1:04:22 Single Cell 5' Gene Expression Output: Demonstrates over 200 million reads greater than 1,000 bp and 2.5 billion reads greater than 400 bp in a one-hour run for polydispersed single-cell RNA libraries.
1:05:10 Long Read Potential: Highlights the capability to generate expandimers up to 2,000 bases, with ongoing efforts to push read lengths further.
1:06:50 Simplex Data Integration: Simplex reads can be easily split into traditional Read 1 and Read 2 FASTQ formats to integrate with existing analysis pipelines like CellRanger.
1:13:06 Innovative Headroom: Roche views this as the beginning of the technology, with continuous opportunities for optimization in accuracy, throughput (e.g., faster pulsing, increased sensor density), and cost efficiency.
1:16:05 Commercialization Strategy (Gustav): Roche aims to develop end-to-end solutions, make SBX an open, research-use-only (RUO) product initially, and minimize the price penalty between large and small studies.
1:18:33 Commercial Timeline: Roche plans to expand its early access program in 2025 and target market availability in 2026.