3 ms·
One thing I'm excited for here is targeted sequencing using read-until. In this method, you monitor the current trace coming off of individual pores, and if you
by rotexo 9y ago
One thing I'm excited for here is targeted sequencing using read-until. In this method, you monitor the current trace coming off of individual pores, and if you determine that the DNA in that pore is not part of your sequence of interest, you can reverse the voltage to remove the DNA and start sequencing another molecule. I think this will open up a lot of applications for human genomics.
If you're interested in Oxford Nanopore, you might also keep an eye out for Roswell Biotechnologies (https://www.genomeweb.com/sequencing/roswell-biotechnologies-harnesses-molecular-electronics-chip-based-dna-sequencing https://www.genomeweb.com/sequencing/roswell-biotechnologies...). TL;DR: their sequencer involves immobilizing polymerases in circuits so you can measure the current changes that occur as the polymerase adds bases to a strand. This might be a good approach, since it doesn't involve optics (like IonTorrent and minION, keeping costs down) and you get a current event per base (as opposed to per 5-6 bases as with minION), making basecalling easier and potentially more accurate. And it will be fast, since you're reading as fast as the polymerase can work. They want to get up to 10kb reads, and I imagine they could increase the consensus accuracy per read by looping around a single molecule of DNA several times, like with PacBio. Seems feasible too (after all, PacBio has successfully been able to integrate polymerases into very small features rather well, so I see no showstoppers there).
- daemonk 9y agoThere is a group in UK (Matt Loose's group) that I think is working a lot on read-until. I haven't really kept up with it. I think the key for read-until to work is ultimately probably going to be better hardware as you need to be able to analyze the trace signal and compare it against a database fast enough to tell the device to kick it out. There are plenty of software optimizations that can be done to compress the signal or extract relevant regions, but the comparison to a database will be difficult to do fast.
- rotexo 9y agoThe computational barriers are definitely an issue. I thought I had read rumors on twitter that ONT was trying to implement read-until on one of their larger-scale sequencers (one of the ones with compute integrated). Barring that, the Cas9-mediated targeting method looks somewhat promising (https://www.youtube.com/watch?v=DGDH-FdoARM https://www.youtube.com/watch?v=DGDH-FdoARM). No need for amplification! Though for applications with human DNA you might run into some issues with the copy number of a sequence of interest if you’re starting out with a normal amount of genomic DNA.
- rotexo 9y agoAlso as a biologist by trade and an audio dsp nut by hobby I loved that they were using algorithms initially developed for music information retrieval to extend the capabilities of a sequencing instrument.
- virusduck 9y agoRead-until has had diminishing utility. Over the last year or so, the speed at which a read can pass through the pore has increased so that by the time you call the bases and determine what it is, the read is likely to have already passed through the pore.