As shown in the above photo, the exhibited instrument consists of a liquid handling robot with eight stations for the Nabsys devices (seen below in close-up). During their presentation, the machine happily whirred away. After the slides, real time data was shown.
Nabsys played it safe with their samples; one was phage lambda DNA and the other a synthetic 20kb standard, and each had a small number of binding sites for Nabsys' proprietary probes. This enabled them to demonstrate generating highly precise consensus maps of the probes, with resolutions approaching 20 bp. But, these are clearly highly artificial test environments which don't really represent actual use cases. Even running one mixture of such defined molecules or better a fragment pool, would have been messier but more in line with the real world.
A number of questions in my showing were around sample preparation and purity. While optical mapping requires very high molecular weight DNA, which requires the tedious in-agarose lysis procedure, Nabsys is confident that standard column preps will be adequate. As far as contamination, the flow cell geometry apparently makes clogging by debris uncommon.
For software, the initial version will generate a map from the Nabsys data and integrate that with sequence data. Future software might more directly integrate the Nabsys data into the assembly process; at issue is whether it is better to try to suggest errors in assemblies generated with existing software or simply use the mapping data to try to get the assembly correct in the first place.
In some later discussion with someone who had attended a different session, the throughput of the instrument seemed huge. At 1 Mbp/s, the needed 20X coverage of even a large bacterial genome would be acquired in a few minutes. In theory, sufficient coverage for a human could be obtained in under a day. But for the moment, Nabsys is targeting sub-100Mbp genomes. So with 8 cells tearing through a microbe every few minutes, a legitimate question is who needs that kind of throughput? Nabsys must be betting that such appetites will emerge.
An interesting future possibility is manually feeding the devices and running them off battery power, which they are specified to do. This could make taking them in the field for, say, foodborne illness studies or other epidemiology, a reality. Ion Torrent has their bus parked on the beach, but with Nabsys might be a scheme that is literally man-portable. At the other extreme would be PacBio, demonstrated to be useful for studying microbes but decidedly not portable. Of course, these three instruments are delivering very different information, so Nabsys would need to prove that their more portable instrument can still deliver valuable information in the context of an epidemic, if they were to pursue such a course.
Nabsys emphasized that this was not a product launch, but an opportunity to get a first look. Undoubtedly, the next phase will be working with scientists with real problems and seeing if positional sequencing can address them. There are certainly problems out there. In addition to my own travails in the field of microbial genome assembly, a number of talks and posters at AGBT have examined issues with repeats and gaps, including a whopping gap in a human mucin gene and the ugly challenge of properly mapping variants within mammalian MHC regions. Many of these posters used either Ilumina/Moleculo or PacBio to attack these problems, though there were also posters from optical mappers OpGen and BioNano Genomics. Nabsys will be trying to elbow their way into the same space, but there will probably be room for a number of players so long as none are a knock-out winner.