This article has been updated to correct that BGI's Copenhagen laboratory currently does not offer commercial sequencing services on MGI platforms.
NEW YORK (GenomeWeb) – MGI Tech, the instrument division of China's BGI, is preparing to enter the North American and European markets with its sequencing platforms and technologies, possibly through a partner in the US. By the end of this year, the firm hopes to be ready to compete with Illumina, which has dominated the global sequencing market for years, on data quality, cost, and service in these territories.
MGI, which has sold 1,000 sequencers in China and the Asia-Pacific region, already has about 20 early-access customers in Europe, has built a manufacturing facility in Latvia, and is assembling commercialization teams in North America and Europe.
At the Advances in Genome Biology and Technology conference in Marco Island, Florida on Saturday, MGI also announced a new sequencing chemistry for its platforms, called CoolNGS, which will go into the hands of early-access users later this year. The new chemistry uses unlabeled nucleotides and fluorescently labeled antibodies to detect base incorporation, enabling longer and more accurate reads at lower cost, according to the firm.
During the meeting, MGI also highlighted its upcoming high-throughput sequencer, MGISEQ-T7, which it announced last October at the ICG-13 conference in China; its MGISEQ-200 and MGISEQ-2000 sequencers; and its sample preparation platforms. It also launched its single-tube long fragment read (stLFR) technology last week, which allows for the assembly of long reads from short reads and is similar in nature to 10x Genomics' linked reads, and it announced a PCR-free library preparation kit for whole-genome sequencing that will be launched in the second quarter. In addition, a researcher from the Wellcome Sanger Institute presented results from a comparison between sequencing data from BGI and Illumina, demonstrating that the two are of comparable quality.
Roy Tan, general manager for MGI Americas, said in an interview that the company is currently assembling commercialization teams in North America and Europe and aims to be ready to sell sequencing instrumentation in these markets by the end of this year. "As soon as we get the team ready, we'll move," he said. MGI is also considering a partnership for its US commercialization and is currently exploring different options, he added, noting that such a deal might take some time.
Because MGI is based in China, the firm started selling instruments in its home country first, where service and support were easier to build, followed by Asia-Pacific. Having established its PCR-free library prep and stLFR technologies, MGI now feels ready to enter Europe and North America, where it will be harder to succeed because of existing competition and high customer expectations, Tan said.
As of now, MGI has no instruments at customer sites in the US, he said, though its San Jose research site – the former Complete Genomics – has several installed. Two MGI instruments are with customers in Canada, with others to follow, and more than 20 customers or early-access partners in Europe have MGI instruments, with the notable exception of the UK, where the firm is hoping to place an instrument soon. For Canada, MGI already has a sales and service team in place, Tan said, which is currently on a three-month training course in China.
In the meantime, US customers can receive MGI sequencing data from BGI Genomics, which offers sequencing services through its Hong Kong laboratory, where several dozen MGI sequencers are installed. BGI Europe's Copenhagen laboratory now also has several MGI sequencing platforms in place, however, these are used for internal training purposes only and not for commercial sequencing services. BGI stopped offering sequencing services through its Philadelphia laboratory a while ago, Tan said.
To prepare for instrument sales in Europe, MGI has built a large manufacturing facility in Latvia, which will become operational by the end of this year and which MGI hopes will shorten delivery times. Right now, exports from China need to go through Hong Kong, Tan said, which is inconvenient. MGI has not yet decided which of its products will be manufactured in Latvia, which Tan said will depend on sales forecasts. MGI currently has no other manufacturing facilities outside of China, where it has two production sites, but it might add more in the future, he said.
While MGI maintains that the comparatively late rollout of its platforms in the US and Europe is mainly related to the need to build sales and service capabilities, various industry experts have said that it is partially due to Illumina's intellectual property around sequencing by synthesis in those regions.
Tan declined to comment on specific intellectual property but said that, in general, MGI scores several factors for making decisions about which territories to enter, and those factors include both regulatory requirements and IP.
At AGBT, Michael Quail, principal scientific manager at the Wellcome Sanger Institute in the UK, presented results from a comparison of Illumina and BGI data, both in a poster presentation and during a talk hosted by MGI.
His team sent microbial DNA samples with varying GC content – ranging from 19 percent to 67 percent – as well as a standard human sample, NA12878, to BGI Genomics for sequencing on the BGISEQ-500 and compared the results to available data generated by Illumina on the MiSeq. For both datasets, a PCR-free library prep was used. Quail also traveled to BGI's Copenhagen laboratory to learn how to do the DNA library preparation and instrument loading.
Overall, he said the quality of the BGI data is very high, with little GC bias, and read coverage was even. The quality of the data was as high as Illumina's but the distribution of errors was a little different. He said sample preparation for the MGI technology is straightforward but took longer than that of competitors, and the MGI instruments appeared to be flexible and user friendly.
Among other results, he found that BGI's read coverage for S. aureus DNA, which only has 32 percent GC content, was more even than Illumina's, which he said might be because Illumina has optimized its technology over the years for GC-rich regions of the human genome at the expense of AT-rich regions.
He also found that BGI's read coverage of chromosomes 6 and 18 of the human sample was more even than Illumina's, covering some regions that Illumina reads did not cover well, though both technologies showed read dropout in the MHC region.
Following an assembly of the human data with the SOAPdenovo assembler, BGI data delivered a "pretty good assembly for a short-read dataset," according to the poster, with an average contig size of 8 kilobases and an N50 of 48 kilobases.
A reference-guided assembly of the BGI and Illumina datasets yielded a larger N50 contig size for Illumina, which also had the largest contig, probably as a result of the longer read length (250 base paired-end MiSeq reads compared to 150 base paired-end BGISEQ-500 reads). However, the average contig size, number of contigs, and chromosome coverage were better for the BGI data, which covered 99 percent of the genome.
The Sanger Institute will likely be interested in taking a look at the MGISEQ-T7 instrument, for which MGI is currently looking for early-access customers, Quail said, and would evaluate it for a range of applications, including not only human genome sequencing but also bacterial sequencing, RNA-seq, and ChIP-seq. However, it would be difficult to maintain two different sequencing technologies in parallel, he said, and to justify the replacement of an existing sequencing technology, a new technology would need to be of better quality and lower cost.
Tan said MGI's platforms will be very cost-competitive with Illumina's. The high-throughput MGISEQ-T7, which will produce between 1 and 6 terabases per one-day run from up to four flow cells using 150 base paired-end reads, has a list price of $1 million and a reagent list price of $5 per gigabase, translating to about $500 in consumables costs per human genome, of which less than 10 percent are related to sample prep. The instrument is expected to go into service work at BGI Genomics later this year, and MGI is looking for early-access customers in different parts of the world. Chinese direct-to-consumer company WeGene has already ordered one of the new instruments.
The MGISEQ-2000 has a list price of $450,000, which Tan claimed is about half that of Illumina's NovaSeq, and a reagent list price of $8 per gigabase, or about $800 per human genome. Its output comes close to that of a NovaSeq with an S1 flow cell, he said. According to MGI marketing material, the MGISEQ-2000 has a maximum data output of 540 gigabases per run using a single flow cell and 1,080 gigabases per run using two flow cells, with 150 base paired-end reads. In addition, 300 base single-end reads are in development.
Finally, the MGISEQ-200 has a list price of about $125,000, which differs slightly by region, and a higher reagent list price per gigabase than the MGISEQ-2000. This instrument is often used for applications such as NIPT that require a lower throughput and are less cost sensitive in terms of cost per base, Tan said. According to MGI, the instrument has a maximum output of 60 gigabases per run from a single flow cell, with 100 base paired-end reads.
Last month, the company also globally launched two automated sample prep systems, which Tan said are available in the US through a distributor. The MGISP-100, which automates library preparation and linear amplification to generate DNA nanoballs (DNB), has a list price of $95,000, and the MGISP-960, which has similar capabilities but provides higher throughput, has a list price of $180,000, which Tan said is 35 percent less than competing platforms. He also mentioned another instrument, MGIDL, which is required for automated sample loading onto the flow cell.
With the company's new CoolNGS sequencing chemistry, which will be compatible with the MGISEQ-200, MGSISEQ-2000, and MGISEQ-T7 and is slated to go to early-access customers later this year, read length and accuracy are expected to increase further and sequencing costs to drop. Tan said the plan is to have the new chemistry replace the current one over time.
According to MGI CSO Rade Drmanac, the current sequencing chemistry relies on stepwise sequencing-by-synthesis (SBS) where 3'-blocked nucleotides are labeled with cleavable fluorescent dyes, which leave behind a molecular "scar" after they are removed. This chemistry is similar to that used by Illumina and others. Instead, MGI has now developed a chemistry that involves unlabeled 3'-blocked nucleotides and uses four types of fluorescently labeled monoclonal antibodies for detection that are each specific for one base. As a result, the nucleotides incorporated into the DNA do not bear a scar. "Now we have our proprietary, better chemistry," Drmanac said. It is unclear, however, whether aspects of this chemistry would be covered by any of Illumina's SBS patents.
The antibodies are not only specific for one of the four bases but also only bind to the nucleotide when its 3'-block is intact. "That's a nice property because as soon as you cleave the block, the antibody is removed in the same step," Drmanac said. In addition, three dye molecules can be coupled to one antibody, which increases their brightness, resulting in a stronger signal and better accuracy.
The fluorescent dyes to label the antibodies are similar to those currently used to label the nucleotides, so no changes to the hardware of the instruments are needed. "We're just changing the chemistry," Drmanac said.
As a result of the new chemistry, read lengths are expected to grow to up to 600 base paired-end reads and 800 base single-end reads, which will increase the throughput of the sequencing platforms. It will also "dramatically improve" the performance of the stLFR technology, Drmanac said, because the longer reads will cover longer stretches of the DNA fragments.
In addition, sequencing costs are expected to decrease further, despite the need to produce antibodies for the new sequencing kits. Drmanac said that antibody technology has developed in recent years, involving bacterial expression vectors now, adding that "we calculated that the cost of antibodies is actually lower than the cost of polymerase."
Several participants of the AGBT meeting who are customers of Illumina said they are happy to see competition from companies like MGI and Genapsys arise, citing worries about Illumina's dominant market position.