Thread by @nanopore

Opening the tech update at the Nanopore Community Meeting, from Oxford @Nanopore, James Clark, VP platform technology #nanoporeconf

JC: a reminder that our goal is to enable the analysis of anything, by anyone, anywhere #nanoporeconf

JC: Looking at about the key qualities of nanopore sequencing: high accuracy data, with high yields, information-rich data, a versatile platform that scales from portable to ultra-high throughput #nanoporeconf

JC: a reminder: this is how nanopore sequencing works. The current trace, called a squiggle, is decoded by basecalling into a sequence vimeo.com/337258910 #nanoporeconf

JC: you can see previous videos, from Oxford Nanopore or the community, at our resource centre nanoporetech.com/resource-centre

JC: this is the range of @nanopore devices – all sizes mean that anyone can access the technology however big the project. PromethION, GridION and MinION Mk1C include embedded compute #nanoporeconf. More here nanoporetech.com/products

JC: PromethION is our largest device, designed for larger samples and larger projects #nanoporeconf

JC: at London Calling we reviewed some PromethION flow cell improvements, in connectivity to the ASIC and microwell structure. More recently we’ve been shipping FCs that include these improvements & are seeing rising data yields from Prom flow cells in the field #nanoporeconf

JC: we perform PromethION benchmarking tests regularly, and in June 2020 we achieved 8Tb in a single run using PCRd DNA, on all 48 flow cells (~170Gb avg per FC). #nanoporeconf

JC: one of the elements of the technology is the membrane that the nanopores are embedded in. We have redesigned a membrane that enables a higher number of pores in channel, which translates to higher throughput. This is now rolled out #nanoporeconf

JC: As a result of these improvements we have a new internal record of *10 Terabases* of data from one run of all 48 PromethION flow cells, averaging 208 Gb per flow cell – with excellent consistency between flow cells #nanoporeconf

JC: the 10Tb record utilises the latest MinKNOW, membrane, and used PCRd human DNA prepped on Hamilton #nanoporeconf

JC: A new ultra-long kit which has been optimised for people interested in those valuable >100kb fragments. Internally we can produce 90-100Gb (read N50>100kb) data from a PromethION FC, with a new record of 4.15Mb single read #nanoporeconf

JC: We are now achieving internally 220Gb per PromethION FC with 10-30kb N50, or 100Gb per FC at 100kb – this requires some interaction during a run. When barcoded, => 2 human genomes at 30X or trio at 20X #nanoporeconf

JC: we're delighted to see users' yields climbing and we look forward to supporting you, building more on this success

https://twitter.com/AW_NGS/status/1331243073553174528

JC: looking at addressing limitations for future yield improvements, we are working on increasing the concentration of mediator in the well and how we can increase channel numbers #nanoporeconf

JC: Moving onto the smallest nanopore flow cell: Flongle. This is designed to enable single, rapid, on-demand analyses at low cost. #nanoporeconf

JC: for Flongle we think about how to drive flow cell costs even lower, to make it possible for more people to use it to run more analyses. This is all about manufacturing innovation: not just flow cell design but better processes for making them at scale #nanoporeconf

JC: Whats the pathway to Flongle being very low cost? Nothing in the flow cell needs to be expensive – here is a prototype single cell channel from back in 2006 in which we successfully ran single pores. #nanoporeconf

Talking of manufacturing innovation - here is the high tech factory we opened in 2019

JC: For low cost Flongle flow cells, we can replace a metal base with a plastic sheet. We can go from lithographic processes to cheaper mouldings. We want to move towards flow cells for just a few dollars #nanoporeconf

JC: We continue to work on voltage sensing; a long running project that is high risk and very high reward – to dramatically increase density of nanopore sensing. Lots of progress and the project is starting to look less risky #nanoporeconf

JC: on voltage sensing chip - if these channels are engineered at a 40um pitch, this has the potential to deliver ~100,000 nanopore channels on a piece of silicon 16x30mm => 3.9Tb / day #nanoporeconf

Thanks to James Clarke for part 1 of today's update - up next, Stuart Reid, VP Development #nanoporeconf

SR: where does the high accuracy of nanopore sequencing data come from? There is a rich signal - patterns to be learned as the DNA travels through the nanopore. A pretty good application of machine learning #nanoporeconf

SR: Different generations of algorithms, on the same chemistry, have moved raw read accuracy of *each single, native molecule* from ~90% to >98% in around 2 years. #nanoporeconf

SR: Research version of bonito, delivering modal single-read accuracy of ~98.3% is available now: github.com/nanoporetech/b… – but it’s ready for prime time – expect to see in Guppy and MinKNOW in coming months #nanoporeconf

SR: high-quality consensus accuracies now achievable with bonito using only half the read depth – R9.4.1 hits Q45 at 30-60X, and R10.3 gets to Q50 at less than 100X #nanoporeconf

SR: combination of Medaka with DeepVariant - similar to community methods - have boosted SNP calling to ~99.92%, expect further improvement with new basecallers #nanoporeconf

SR: new SV calling pipeline makes uses of excellent Community tools LRA @mjpchaisson and cuteSV to produce high-quality calls twice as fast and with half as much data #nanoporeconf

SR: new ultra-long kit achieves 90-100 Gb of data per PromethION, with hundreds of thousands of reads >100 kb, and hundreds of reads >1 Mb; will be released shortly. Ideal for spanning complex SVs, spanning repeats, resolving regions challenging to assemble @circulomics

SR: re-basecalling T2T consortium CHM13 data with bonito gives 80 Mb N50, Q46, BUSCO 96%. Pore-C can improve further with long-range contact information – see Poster section for more #nanoporeconf

SR: new PCR cDNA kit in early access now; improvements give ~100 million reads per PromethION flow cell with >85% full-length & mapped, to deliver unprecedented ability to quantify isoform expression levels #nanoporeconf

SR: reminder – only @nanopore can directly examine modified bases, with no GC bias, long read lengths, phasing etc – it’s all in the data. R10.3 shows high sensitivity to modifications, delivering even more accurate per-read methylation calling #nanoporeconf

SR: R9.4.1 is good at methylation calling - an F1 score of 91.5% per read - but R10.3 is an improvement at 95% #nanoporeconf

SR reviews current levels of different types of accuracy - more information can be found here nanoporetech.com/accuracy #nanoporeconf

SR: you don't necessarily need to basecall to do Adaptive Sampling. A user can programme their system to accept or reject strands based on a configuration specified in software, enabling rapid sequencing of regions of interest, no sample prep

SR: a grand challenge @nanopore: make bioinformatics easy. EPI2ME Labs is the simplest way to explore bioinformatics with nanopore #nanoporeconf

SR: The MinKNOW App is now available for download on iOS and Android devices, more information and download links here: nanoporetech.com/about-us/news/… #nanoporeconf

SR: This week Oxford Nanopore has generated modal *raw-read accuracy* of 99.1% (99%=Q20) using a new chemistry with Bonito, delivered on internal validation sets with a substantial fraction of these raw reads above Q20 #nanoporeconf

Next up @RosemaryDokos who is going to update us on the usability and accessibility of our technology; critical to support the goal of enabling the analysis of anything, by anyone, anywhere #nanoporeconf

RD: We focus on all parts of the workflow – versatile and easy sample prep, sequencing and analysis, including open communities #nanoporeconf

RD: on preparation, its important to have the right tools. We have native DNA preparation methods and will be adding the ultra-long kit to this. And there are extensive options on the amplified side, & direct RNA – as well as ways of targeting regions of interest #nanoporeconf

RD: we wrap the preparation methods with protocols, and have just updated the knowledge section of the nanopore community. Please give us feedback on the new protocol layouts #nanoporeconf
community.nanoporetech.com/docs/

RD: EPI2ME labs has notebooks explaining how to process and manipulate nanopore data. GitHub has all our pipelines in open source. EPI2ME labs has now been available for six months #nanoporeconf

RD: over the coming months we’ll be working to take the workflows from EPI2ME labs into EPI2ME, so we can work towards a single click bioinformatics solution #nanoporeconf

RD celebrates the new working environments for biologists #nanoporeconf

RD: Oxford Nanopore is collaborating with Hamilton to automate high throughput library preparation workflows on Hamilton’s NGS STAR 96. #nanoporeconf

RD: on Hamilton prep, 1st protocol enables preparation of 96 samples in under 5 hours, with only one hour of hands-on time. One run fills two PromethION 48s, enabling high yield, highly validated sample prep and sequencing output. #nanoporeconf

RD: we’re also working with Opentrons to deliver mid-range lab automation suitable for MinION/GridION. The robots are accessible, compact, & can support 2,500 LamPORE tests/day or automation of the Ligation kit, ARTIC native barcoding, & suitable for future PCR based applications

RD: we can’t talk about automation without mentioning VolTRAX – key to accessibility

RD: we’ve been completing all the validations for the PCR ready cartridge, including enabling RT-PCR, and are deploying it for ARTIC whole genome sequencing of #SARSCoV2; prepare on VolTRAX, sequencing on Flongle for $75 #nanoporeconf

RD: whats available in store for VolTRAX #nanoporeconf

RD: Three configurations of GridION are now available, for research, Q-line for applied markets, and CE-IVD for diagnostic use #nanoporeconf

RD: We make devices available with no capital requirements. As soon as you invest in us, we invest back, with a seamless upgrade pathway and no hefty renewal fees. The same users who were generating 1Gb of data on MinION a few years ago are now generating tens of Gb #nanoporeconf

RD: the full range of specs of nanopore products nanoporetech.com/products/compa…

RD: with high throughput projects on PromethION, there are choices of workflows, all delivering cost competitive, high value genomes #nanoporeconf

RD: PromethION’s capital-free model delivers not only value, but huge capacity to deliver whatever volumes of data are required #nanoporeconf

RD: Nanopore data is information-rich, so the data that you generate has additional value #nanoporeconf

To see the whole talk from @RosemaryDokos, James Clarke and Stu Reid’s, visit #nanoporeconf (register nanoporetech.com/ncm20 if you aren't in the conference platform already) to watch the talk on demand and we'll make it available at our website soon #nanoporeconf

@threadreaderapp unroll!