Monday morning thought for y'all...

No one innovates & disrupts by accepting the status quo, they do it by burning the text books & giving it a go! 

I had a crazy idea in my garage once and just thought I'd give it a go. It worked and now I have the coolest company in the world, QuantuMDx.

Cancellation of the Archon X-Prize for Genomics

I appear to be on a rant roll… 

Why, oh why did the folk organising the Archon X-Prize cancel it? Rather than cancel it, they should have removed the entry fee. Only companies selling sequencing devices and services today can spare important R&D dollars for the entry fee. Unfortunately these companies technologies will never be able to even come close to achieving the criteria set to win the prize. 

Indeed, the criteria were set, rightly, to a very high level that only 4^th or 5^th generation, totally outside-of-the-box-out-the-door-down-the-road-over-the-lake-and-into-the-next-state technology will achieve. Therefore, rather than close the door on the prize, it should have been extended to allow for the development of this technology. It is not our (the next generation of sequencing rockstars) fault that the present old guard have limited their thinking and have settled with what they have got. 

Out of stagnation comes innovation… so reinstate it!!! 

The Wonderful World of NGS (The Rant) - Part 2

And the rant continues...

The long(er) reads from Pac.Bio’s sequencer have shown the sequencing community (the 'Seqosphere' - I like dumb names, accept it) that long, single molecules reads must be the future of discovery genomics. Bravo Pac.Bio! However, whilst an awesome technology, some of the earlier chemistries demonstrated to the world that relying on biology and fluorescence have their limitations. Recent patents and grant applications suggest a move to the dark side for Pac.Bio. If only they could find a way to get rid of the enzymes, reduce the device costs & size, the bioinformatics burden and run times, then they’d be laughing.

The present excitement in the field, as we look to the future, is all centered on nanopore technology, which is, so far, promising to deliver a 'dream machine'. However, Oxford Nanopore Technologies (ONT) are apparently still relying on biological (lipid) membranes and proteins (reading a sliding window of 3-5bps) which is always going to give you stability issues and errors. Nabsys appear to have given up on nucleotide sequencing for now in favor of looking at structural elements (sounds a bit like the US Genomics story all over again) and Genia, well who knows where they are at other than they are able to make great microelectronic solutions. 

Yet, I applaud them all as they understand that the next generation needs to be another major step-change in sequencing. I further ‘doff my cap’ in their general direction, as what they are attempting is technically challenging, to the point of being close to impossible.

The concept of nanopores is a good one, direct read of single molecules over huge lengths. But the basic fundamentals of nanopore sequencing worries me somewhat (of course I’d say that, I’m not developing a nanopore sequencer, so am completely bias). The resolution isn’t there for true single base pair identification (as I mentioned above) and therefore the accuracy will most likely drop again. Moving to a solid state Graphene nanopore can theoretically solve this, but how long will this take to develop (recent noises from ONT suggest this is one research route the are pursuing at their base in Cambridge, MA). More importantly, there is only one detector, i.e. one shot at base calling (OK, two shots if you form a hair pin loop in the dsDNA and linearize it, or more if you oscillate it, which doesn’t necessarily get rid of the systematic errors associated with the method of base calling), so you’ll still be relying on sequencing many molecules to get the accuracy level right; ONTs ‘run until…’ approach.  

However, nanopore sequencing IS a paradigm shift and it IS truly stunning and we should celebrate companies like ONT and support them in any way we can! They are fighting enough battles with the technology development and the intense pressures venture capitalists give a company, they don’t need us, the Seqosphere, ridiculing them for delays that are normal for any hi-tech development project. Let us not forget Solexa telling us all in 2006, that their technology would sequence a human genome before the end of the year… "he just didn’t say which year" (as Harold Swerdlow famously said). Same thing is happening now with ONT. I think we’ll all agree that Solexa’s technology has done pretty well since finally sequencing its first human genome (well enough of it for people to say it was fully sequenced... it wasn't complete, no human genome has been fully sequenced).

When thinking about DNA sequencing using nanopores, I find myself wishing we could stack thousands of different nanopores on top of each other and have one DNA molecule translocate through the lot, with the sequence read as it passes through each nanopore. Why sequence thousands of molecules one time when you can sequence one molecule a thousand times? It stands to reason that a LastGen 2.0 technology (I call it this as Stefan Roever, CEO of Genia, once claimed their nanopore technology would be ‘LastGen’ Sequencing) would be able to do this, to directly read a single and whole chromosome thousands of times, in different ways in a single run. This will limit stochastic errors and weed out systematic errors. Imagine how important this tool will be for single cell applications, such as surveying the genomes of different cells within a tumor for instance.

Is this possible, sure, of course, it just needs some out of the box thinking, some money and a bunch of smart people to build it… it’s one of our internal projects at QuantuMDx, but surely we’re not the only group looking to develop the next gen, of the next gen, of the next gen… of the next gen of sequencers! We are going through the early stages of creating a novel technology for the field and it’s daunting. Our field is unforgiving of normal developmental hiccups such us early prototype error rates and delays in product launches. This resistance and sometimes mockery filters to the investment communities which limits companies abilities to raise the needed funds to make all our dreams a reality with the next big killer device.

So my call to arms is this; to all young scientists, senior scientists, P.I.s, institute heads and leaders of the bioscience industry, don’t accept the technological status quo, the bottom-line-driven fear mongering rhetoric of large sequencing companies who want you to believe their technology is the best and won’t be beaten whilst churning out diatribe to derail new entrants into the field. Do accept and understand that the funding-attracting sensationalism from fledgling companies claiming the $100 dollar genome with their “Last-Gen” killer technology is a necessary evil to lure investment dollars to produce killer new tools. Accept that the development of the next blockbuster DNA sequencing tool is hard, super hard, and that delays are inevitable and shouldn’t be mocked. Accept that venture capitalists insist on the dreaded ‘stealth mode’ to protect their investment from copycats. Don’t dismiss novel approaches and companies because they are not run by people with a track record in making sequencing tools, or you can’t immediately see how it can work. 

Dream of a world of high quality super long reads of single, whole chromosomes, telomere to telomere in a single rapid run, read out by cheap re-usable chips, Gattaca style. Help, advise and support these crazy innovators to make cheaper, faster, better sequencing technologies, because one of them may well deliver and the wonderful world of NGS will suddenly become even more fabulous!

The Wonderful World of NGS (The Rant) - Part 1

A thousand apologies for the month between posts, however I have been bouncing around different cities speaking at conferences and meeting potential investors for QuantuMDx (I shall discuss this in a later post).  During my travels I have spent a lot of time thinking about DNA sequencing technology and why I am so frustrated with the field in general. 

So I thought I would treat you all to a rant, lucky you…

Highly parallel shotgun sequencing (i.e. breaking up the chromosomes in to short fragments and sequencing, from either end, 25-400bp+ nucleotides, before assembling all the short reads into contiguous sequences and then ordering the contigs into a scaffold along a reference sequence) is great, but can never provide a true and full de novo genome sequence, which is required for the next phase of discovery genomics. It should instead be focused to perform targeted re-sequencing for clinical applications and newer technologies developed and optimized for discovery.

Short read parallel targeted sequencing tools are OK for exome sequencing, as there aren’t many repeats in coding regions and as ~80% of all [inherited] disease-causing mutations are located within protein coding regions, many researchers and clinicians appear happy with this. However, for discovery, the fact that we are unable to resolve ~20% of disease causing variation, means it's not the tool for discovery (I am sure this is not news to most of you) and in the future, not even the tool for the clinic. What about structural variation, repeats, CNVs, centromeric and telomeric regions. These are important and as we’ve been accepting mediocrity in our sequencing technology for so long, we know very little about them, as we’ve been unable to sequence them.

Sure highly parallel shotgun sequencing can get us so far with genome sequencing and will certainly maintain it’s position as the dominant force in ‘whole’ (note: we have still not sequenced a complete human genome, nor are we able too with presently available technology) genome sequencing, blah blah blah. However, it’s not as great as we all thought it was when 454 & Solexa released their initial visions. We (genome scientists) deserve better than that, we’re bloody smart people, yet here we are accepting imperfect technology.

Short reads suck, assembling to a reference genome sucks more, missing out on important structural variation is scandalous. Why are we investing so much time and effort into making our sequencing and bioinformatics tools incrementally better? Isn’t that just accepting the status quo, essentially telling all budding young scientists that we’re giving up on technological innovation so that we can polish a turd…. C’mon people wake up and smell the asparagus (a random ‘23&me’ reference, if you’re now scratching your head and thinking I’ve lost the plot)?!?!

Yes, I'm spouting histrionics to make a point, shoot me!

It is the technology that's lacking at the moment. So we should focus on ways to build better devices that can read long fragments of single DNA molecules (preferably whole chromosomes, from telomere to telomere), directly, without relying on error prone enzymes, or reporter moieties and labels (move away from the ‘light’ people and join the dark-side… pH too for that matter) and with significant redundancy in the system to weed out stochastic and systematic errors!

These technologies exist today; they exist in the heads of scientists and in the labs of small companies vying for investment dollars. So why don’t we invest our time, money and efforts in helping & supporting them rather than drinking the shotgun-Kool-Aid and deriding the brave souls attempting to make the dream sequencer.