Wednesday, September 21, 2022

Notes From Coffee With MGI

A couple of weeks ago  I sat down for coffee with a pair of MGI representatives - American Region CEO Yongwei Zhang and Director, Global Business Development Damon Zhang. Since I hadn’t been at AGBT 2022 (my 2023 application already filed!). Yongwei and I had planned to try to catch up the next time he was in Boston area, so I braved our current subway issues (not one, but two major lines shut for extended maintenance!) and covered a range of topics.

MGI should be interesting to follow. Due to a protracted patent battle, they have been selling all over world except in U.S. or some countries in Europe. But after victories over Illumina and a litigation armistice to last three years, they are poised to enter the U.S. market. They have a full array of sequencers and other hardware to go with them, which should make them a formidable contender. They also just landed a $500M warchest via an IPO on the Shanghai exchange. But they also recognize that there are many perceptions around the company that may hinder their ability to sell into the U.S. market. And MGI just announced an addition to their benchtop line with the DNBSEQ-G99, to actually be available "overseas starting in 2023.

On the tech side, coming soon will be the CoolMPS chemistry, a very interesting application of

antibodies to distinguish between four different reversible terminators. Come the new year, their standard MPS chemistry will also be marketed in the US. While CoolMPS is serving as a bridge to an earlier market entry, Yongwei stressed that this is mostly because the existing instruments have been optimized for standard MPS. He promised that sometime next year the unique capabilities of CoolMPS will be demonstrated on a new instrument, but no further details on that point. He also said there will be a major announcement next month at ASHG, with “highest throughput” instrument being officially launched in US.

Yongwei wanted to emphasize that MGI is focused on providing complete sequencing solutions to customers – DNA extraction and library prep automation, sequencing itself and IT solutions. He contrasted their range of front end automation with other short read competitors’ lack of such offerings. Indeed, Illumina’s main foray in this space was the failure known as NeoPrep, which was a very low throughput device even if it had been reliable.

He also stressed that MGI has no interest in building sequencing applications that might compete with customers.  I would note that contrasts to Illumina’s pattern of watching customers build applications such as non-invasive pre-natal screening or oncology screening and then leaping in with their own offerings.

Data security is important to many in the sequencing market, and MGI believes they are viewed with suspicion on this point. Yongwei emphasized that it is possible to run both their sequencer and their computational environment entirely air-gapped from any external network. If users are comfortable, they can choose to grant greater access outside the lab, there are varying levels, ranging from allowing instrument monitoring for service purposes to off-site informatics.

I asked how they would approach a customer who already had a strong Illumina operation. Starting at process end, Yongwei pointed out that in the end they produce FASTQ and the error profiles are very similar to Illumina, so there is very little on the downstream informatics side that must change. Loading the instrument is similar, except the DNA nanoball creation step (analog of clustering) occurs off-instrument in a short reaction rather than on-instrument. MGI’s DNBSEQ-G400 flowcells allow four independent lanes to be loaded without any accessory hardware and instruments can run two flowcells completely independently. Upstream of clustering, library prep and sample prep are essentially the same.

On the pricing front, MGI’s goal is to be “lower than any competitor” – which is interesting given the suddenly hot competitive mix in the field of Element now and Ultima and PacBio short reads next year - and maybe even Singular. Of course, the cost calculation is complex and some players such as Element and Singular are already differentiating their consumables for counting vs. non-counting applications.

Yongwei also keyed me in to the instrument nomenclature, and it’s simple. G is for Gigabase and T for terabase – each instrument is coded by the targeted daily productivity. The C4 single cell technology appears to be slated to roll out in the US later than the sequencers. I posed the idea that C4 might be a way to pull customers to the platform rather than the model of sell sequencers first, sell single cell later; Yongwei politely thanked me for the suggestion - I’m not expecting him to change course on my whim!

Yongwei also outlined the corporate structure of the BGI/MGI group for me. MGI is focused on hardware and software tools, BGI Genomics provides outsourced sequencing from their facility in Hong Kong and BGI Research is incubating advanced genomics technology such as BGI’s Stereo-Seq spatial sequencing technology. For the foreseeable future this technology, which can map transcripts on a full section of a mouse embryo, will be accessible only via collaboration with BGI Research. One challenge with such large sections is limited availability of devices which can prepare the appropriate thin sections - with many extraordinary technological advances the supporting technology brings up the rear! Even so, there are already a fleet of Stereo-Seq publications covering  cervical cancer,  mouse (brain, lung, embryos ), axlotl brain, zebrafish embryogenesis, Arabidopsis, and peanuts.

MGI entering the US market is just one piece of a very sudden transition of the sequencing market from near total dominance by Illumina to what promises to be a rollicking marketplace, with the sequencing community likely to benefit from competitive pricing and further innovation.


Anonymous said...

Hard agree, single cell transcriptomics through current market sucks, I'm more willing to do the original digital droplet method than pay for other systems right now, but I think I'll be skipping this tech all together, it's like 1 step over sequencing by synthesis but several steps behind nanopore methods

James hadfield said...

Nice summary Keith. The single cell opportunity is huge as costs are high for both sample prep AND sequencing. Looking forward to hearing more.