Open Benchmarking of CycloneSeq™ for Complete Bacterial Genomes
Benchmark Data and Analysis of New CycloneSEQ Using Novel Nanopore Sequencing Technology Demonstrates Ability to Sequence Complete Bacterial Genomes
Following on from the recent official launch of BGI's new CycloneSEQ™ is the first independently reviewed benchmarking data. The CycloneSEQ™ platform delivers long-reads using novel nanopore technology. This study tests the performance of a new platform in sequencing diverse microbial genomes and presents both the raw and processed data to allow others to scrutinize and verify the work. To provide further transparency, the peer-reviews, scripts and protocols of this benchmarking process are also included, allowing others to directly assess this new technology. By directly reading DNA molecules without fragmentation, the analysis demonstrates that CycloneSEQ delivers long-read data with impressive length and accuracy, unlocking gaps that short-read technologies alone cannot bridge. This work has been published in the open-science journal GigaByte.
The Challenge: Filling the Gaps in Millions of Bacterial Genomes
While long established short-read sequencing technology remains cost-effective and accurate, struggles to assemble complete, circular bacterial genomes remain—leaving critical gaps in our understanding of microbial functions. The researchers from BGI-Research have combined CycloneSEQ long-reads with DNBSEQ™ short-reads, achieving closed, circular, high-accuracy genomes for common gut bacteria.
As a standard validation approach, the researchers sequenced the bacterial reference strain Akkermansia muciniphila ATCC BAA-835, yielding 12.07 Gbp of long-reads with an average length of 11.6 kbp. The hybrid assembly method outperformed both short- and long-read-only approaches, successfully assembling a complete genome of ATCC BAA-835 with a mismatch rate lower than 0.0001%. The researchers subsequently sequenced 10 common strains isolated from the human gut and successfully closed all 10 bacterial genomes, as well as tiny phage/plasmid circles typically missed by short reads alone. Using long-read-only assemblies, they managed to put together complete, circularized genomes for 8 out of 10 strains, whereas using the previous generation short-read-only assembly failed to produce a single complete genome. The researchers also revealed that the issue was not the inability of short-read sequencing to detect relevant regions but rather the inherent limitations of short-read assembly, typically caused by the presence of repetitive regions and high GC content in these genomes
The technology also was tested using complex microbial communities. In a 21-strain synthetic gut community (18 bacteria, 2 fungi, 1 archaeon strains), hybrid assembly using CycloneSEQ + DNBSEQ outperformed some other single-method approaches, yielding 5 complete metagenome-assembled genomes (MAGs)—which was not achieved by short- or long-read assemblies alone.
CycloneSEQ's long-reads currently deliver sufficient length and quality for circular genome assembly, but the using short-reads remains essential for polishing accuracy. The researchers future work is aimed assessing and using non-synthetic samples, fine-tuning the balance between short-read and long-read data to provide even faster, higher-quality assemblies.
Additional Reading
Citation:
Hewei L, et al. Efficiently Constructing Complete Genomes with CycloneSEQ to Fill Gaps in Bacterial Draft Assemblies. GigaByte. 2025. https://doi.org/10.46471/gigabyte.154
BioRxiv preprint version: https://doi.org/10.1101/2024.09.05.611410
Data:
Liang H; Zou Y; Wang M; Hu T; Wang H; He W; Ju Y; Guo R; Chen J; Guo F; Zeng T; Dong Y; Zhang Y; Wang B; Liu C; Jin X; Zhang W; Xu X; Xiao L (2025): Supporting data for "Efficiently Constructing Complete Genomes with CycloneSEQ to Fill Gaps in Bacterial Draft Assemblies" GigaScience Database. https://doi.org/10.5524/102694
Protocols.io protocol collection:
Hewei L, Wang M. CycloneSEQ Library Construction and Sequencing Protocol for Isolated Bacteria. protocols.io. 2025. https://dx.doi.org/10.17504/protocols.io.kqdg3k3kev25/v1
