This presentation details the Hartwig Medical Foundation's evaluation of RO's new SBX sequencing technology as a promising alternative for rapid whole genome sequencing (WGS) based cancer diagnostics. The foundation, a non-profit performing routine WGS for Dutch hospitals and developing open-source analysis software, aims to address existing challenges of high cost and long turnaround times.
A beta test involving 30 tumor-normal patient pairs and controls demonstrated the SBX platform's capabilities. While initial raw data showed some GC bias differences (SBX performing better at high GC, Illumina at low GC) and mapability variations due to read length, overall coverage in high-confidence genomic regions was comparable to their current Illumina NovaSeq 6000 production setup. The foundation adapted its platform-agnostic, ISO-accredited analysis pipeline, which empirically determines error rates for each sample, to process SBX data. This revealed lower empirical SNV error rates for SBX reads (Q50 for duplex vs. Q30 for simplex). High concordance was observed for both single nucleotide variants (SNVs) and indels between SBX and Illumina, with private variants generally being low-frequency or related to coverage differences in challenging regions like homopolymers. Preliminary analysis of structural variants and copy number alterations also showed promising similarity. Crucially, the study concluded that observed platform differences had no clinical impact on therapy guidance for cancer patients.
Evaluating the RO SBX Sequencing Platform for Rapid Cancer Diagnostics
0:03 Hartwig Medical Foundation Mission: A non-profit performing routine whole genome sequencing (WGS) for cancer diagnostics in Dutch hospitals, developing open-source analysis software to drive data-driven decision support tools, and building a "learn and care" cycle to improve future patient care.
0:15 ISO Accreditation & Turnaround Time: The entire WGS process is ISO accredited with an average turnaround time of six days.
0:23 Key Challenges: Routine implementation of WGS is hampered by high costs and long turnaround times. The foundation actively explores new technologies to address these issues.
0:44 RO SBX Beta Test: Hartwig partnered with RO to evaluate their new sequencing technology. The prototype equipment was installed in September 2024, with initial production runs in December and the presented data generated at the end of January.
2:31 Experimental Setup: The study analyzed 30 tumor-normal patient pairs and two E. coli cell controls and duplicates, totaling over 120 30x genome equivalents. This was spread over eight 4-hour runs on a single chip, conducted manually initially but later automated on a Biomek platform.
4:02 Raw Data & Coverage: On average, 150 gigabases of data were obtained for normal samples and 410 gigabases for tumor samples. Full duplex rate was about 70%, with an N50 of ~190 base pairs for full duplexes and ~110 base pairs for partial duplexes.
5:07 Mapped Coverage Comparison: Mapped coverage averaged 105x for tumor and 47x for normal, comparable to their Illumina NovaSeq 6000 production data for the same patients.
5:40 GC Bias Differences: SBX showed coverage drop-off earlier at low GC content compared to Illumina, but demonstrated much better performance at high GC regions.
6:19 Mapability & High-Confidence Regions: Due to shorter read lengths compared to Illumina (300bp vs. 2x150bp paired-end reads), SBX had lower mapability in some repetitive regions. However, within genome-in-a-bottle high-confidence regions, both platforms showed almost identical coverage performance.
7:16 Variant Calling with Redux: The foundation uses its custom, platform-agnostic software package, with a key component called Redux, which empirically determines error rates specific to each sample and platform, enabling high-sensitivity somatic SNV and indel detection without modifying downstream tools.
8:56 Empirical SNV Error Rates: Redux determined a Phred score of 30 for simplex SBX reads and 50 for duplex SBX reads. The empirical platform SNV error rates for SBX were found to be lower than on the Illumina platform.
9:43 SNV & Indel Concordance: High concordance (shared variants) was observed for SNVs across E. coli samples and 30 patient samples, with very low rates of private variants. Private variants were typically low variant allele frequency, or due to no coverage in one platform, or in specific challenging regions like T-homopolymers for indels (potentially PCR artifacts).
12:06 Structural Variants & Copy Number: Preliminary analysis showed highly similar copy number profiles between platforms, and structural variant calling looked promising, though further tool optimization is needed.
12:45 No Clinical Impact: Critically, despite minor platform differences, no clinical impact was observed for cancer patients, meaning therapy guidance remained consistent between technologies.
13:00 Conclusion: The SBX platform provides a very promising alternative for rapid WGS-based cancer diagnostics.