SHENZHEN (GenomeWeb) – BGI Group subsidiary MGI launched a new high-throughput sequencing instrument at its 13th International Conference on Genomics (ICG-13) here this week.
In a presentation, Jian Liu, executive vice president of MGI, said that the instrument uses the same DNA nanoball sequencing technology that underlie its other platforms but will have much higher throughput, between one and six terabases per day, or up to 20,000 genomes per year. The MGISEQ-T7 has four flow cells that can run simultaneously and independently of each other, and can each run entirely different applications. A human genome can be sequenced on the instrument in less than 24 hours, using 150-base paired-end reads.
In a follow-up interview, Hui Jiang, chief operating officer of MGI, said that the MGI team focused on upgrading the optical system to make it faster and higher throughput than that of the firm's previous instrument, the MGISEQ-2000. In addition, she said, the new optical system enables the DNA nanoballs to be more densely configured, which also boosts throughput. The flow cell itself was redesigned to improve the density and the researchers developed a new basecalling software to better recognize the signals from the bases.
MGI plans to make the system available to early-access customers in the second quarter of 2019, with a full commercial launch in the third quarter of 2019. The firm declined to disclose a list price for the instrument but said that sequencing reagent costs will be close to $100 per genome.
The platform's specs are reminiscent of the Revolocity, a system that BGI announced in 2015 but then halted a few months later. Jiang acknowledged that while the two platforms do share some similarities — both were designed to be very high throughput and to run multiple flow cells independently — the major difference is in the design of the system.
The Revolocity involved a robot in the center of several modular components, which moved samples through the process, and the system had a footprint of 1,500 square feet. By contrast, the MGISEQ-T7 is a self-contained instrument about the size of a very large refrigerator, more amenable for production and for users to install in their labs. The MGISEQ-T7 is a "closed system" and "more user friendly," Jiang said.
Along with the sequencing instrument, MGI also launched a new sample prep system, the MGISP-960, an automated platform that includes liquid handling and PCR and is designed to prepare 96 samples at once. The system is currently available for purchase and Jiang said it could be shipped about one month from order time.
She noted that the first customers of the MGISEQ-T7 will most likely be based in China, but MGI plans to expand to the broader Asia Pacific region and to Europe, and potentially eventually to the US.
MGI also plans to apply for regulatory clearance for the system from the China Food and Drug Administration. Jiang said that although initially, the instrument would not be used for clinical testing, having CFDA approval would help drive down the cost of clinical testing, a particularly important consideration for the Chinese market.
BGI Group now offers a variety of sequencing platforms. Those classified under the BGISEQ brand are primarily the instruments that BGI runs in house for service work, while the MGISEQ-branded instruments are the ones it sells externally. They all use the same DNA nanoball sequencing technology.
In terms of performance, Liu said in his presentation that the MGISEQ-T7 will have an accuracy of 99.4 percent.
That's comparable to the performance of other BGI instruments, the MGISEQ-2000 and the BGISEQ-500, as reported at the American Society of Human Genetics annual meeting last week. Chris Mason, an associate professor at Weill Cornell Medicine, discussed assessing the instruments using several Genome in a Bottle reference samples. BGI performed the sequencing and Mason's lab conducted the bioinformatics analysis for the study.
For library preparation, the study used the PCR-free whole-genome sequencing protocol on the MGISEQ-2000 and BGISEQ-500, as well as a protocol known as single-tube long fragment read (stLFR), which BGI researchers described in a BioRxiv preprint in May, on the MGISEQ-2000.
Mason noted that the data quality was high, with Q values generally above 35 for all protocols across the paired-end reads. There was some GC bias, at the very far ends of high and low GC composition.
For the data analysis, his team used Sentieon, a set of software tools that is similar to the GATK toolkit but operates faster.
With the PCR-free protocols, sensitivity was above 99 percent for SNP calling and between 95 and 97 percent for calling indels. For the stLFR protocol, sensitivity was also above 99 percent for SNPs and about 97 percent for indels. The researchers compared the stLFR data to data generated on the 10x Genomics platform and found that the stLFR protocol generated fewer false positives for SNPs and indels, as well as fewer false negatives for indels. Both platforms had similar high sensitivity for SNPs but the stLFR protocol had higher sensitivity for indels. Interestingly, though, when the researchers analyzed the variants that the two platforms detected, there was not a lot of overlap. "We can't say what's driving the differences," Mason said, adding that he is "not sure which [platform] is correct." Many of the discordant variants seem to be near centromeres and telomeres, which are highly repetitive, he added.
Radoje Drmanac, chief scientific officer of Complete Genomics and senior vice president at BGI, declined to comment on the stLFR preprint until it is published in a peer-reviewed journal but said that the stLFR kit is now available for early-access use. Ultimately, he said, the kit will be made to work on any sequencing platform, not just the MGI and BGI instruments.
Andreas Keller, chair of clinical bioinformatics at Saarland University in Germany, who previously compared data from the BGISEQ-500 instrument with Illumina HiSeq data as well as microarray data for profiling microRNA, said on the sidelines of the ICG-13 conference that his lab has now purchased two BGISEQ-500 instruments and has been very happy with the results.
"It's high-quality data with short turnaround times and low error rates," he said.
Keller said that the technology is well suited for the needs of his lab, which focuses primarily on small noncoding RNA. In particular, he said, the technology has shown less bias than he previously found with Illumina sequencing, which might be attributable to the fact that the DNA nanoball technology uses linear amplification as opposed to standard PCR. That results in less amplification bias, which he noted is especially important for his application of quantifying RNA molecules.
The big question is whether BGI's sequencing technology will ultimately be able to compete with Illumina's. Keller said that with regards to the MGISEQ-T7, if the specs for throughput and accuracy are as described by BGI, the system could be a "game changer," since it would be on par with Illumina's NovaSeq.
BGI could become a competitor to Illumina, and "that's good for the market," Keller said.
Mason added that the technology is "worth keeping an eye on," although he said it's too early for him to consider replacing the Illumina technology that he is using for current projects.
BGI definitely has a ways to go in order to compete with Illumina, but MGI's Liu noted that the company has so far installed 950 instruments in 12 countries, though predominantly in China.
Another user of the technology, Kedar Natarajan, assistant professor at the University of Southern Denmark, has tested the BGISEQ-500 for single-cell RNA sequencing. During a presentation at ICG-13, he described a previous study his team performed when he was a postdoctoral researcher at the Wellcome Sanger Institute and the European Bioinformatics Institute to survey various single-cell RNA sequencing technologies. That work was published in Nature Methods last year.
Now, he said, he is expanding on that work to compare the BGISEQ-500 and Illumina HiSeq platforms for single-cell RNA sequencing. For that study, Natarajan's group sent samples to BGI for sequencing on the BGISEQ-500.
The motivation for the study was to see whether different types of library preps and sequencing technologies would impact results, he said. The BGISEQ technology, for example, involves circularizing cDNA. Results are so far preliminary, he said, but are generally comparable between the technologies. Natarajan added that the group plans to submit the study to a peer-reviewed journal for publication.