Exponential technologies and electronic DNA detection

We are living in a world of exponential technologies, and genomics is one of several where expectations are not set at something 30% better or cheaper or faster, but 300% within the space of a few years. The term ‘exponential’ is starting to gain traction in the general public; here in the Washington DC area a financial adviser has a radio talk-show on the weekends, and at times will swerve into this topic as an investment category (as well as promote a book he wrote).

Moore’s Law

Of course Moore’s Law has been a generational shift in how technology can quickly overwhelm even the simplest expectations. A few months ago, while reading the novel ‘Ready, Player One’ I thought about all the enjoyable hours of my teenage youth at the local Venice brewpub putting dimes (then quarters) into the videogames there. The brewpub is long gone (that stretch of Main Street in Venice is now quite a site to behold), the videogames only a collective memory (although alive again in the Retropie world), and kids today can recognize hints of those 8-bit games while exploring MineCraft. How quickly we went from 8-bit Space Invaders and Apple ][ computers with 48K of RAM, to photorealistic Star Wars Battlefront and a smartphone in my pocket with 16GB of RAM.

For those who bother to check specifications, the distance between an Atari 2600 (introduced in 1977) to a modern XBox forty years later is about 1700 times as fast in terms of clock speed of the microprocessor, not counting the bit depth (i.e. 8-bit vs a 64-bit one in the XBox). And depending on how you would equate clock-speed x bit depth, a simple multiplication gives you 13,600 times the speed. We have become so accustomed to this kind of technological progress we don’t often stop to think about it: what else has improved some 13,600 times as much?

The Pony Express

For fun let’s think perhaps of a letter going across the country by horse: the Pony Express made the trip of 2,000 miles in 10 to 13 days. Using 12 days, that is 166 miles / 24 hours, or an average of 6.9 miles / hour in 1860. A modern jet plane, over one hundred years later, traveling from St. Louis to San Francisco (a flight distance of some 1,760 miles, but there are no direct flights from Cincinnati to San Francisco I just learned!) will take an estimated 4 hours 21 minutes on United. That is an average speed of 409 miles / hour, or an improvement over the Pony Express of a factor of 59-fold.

Fifty-nine fold increase in transport speed over the course of 100+ years. A thirteen-thousand fold increase in 40.

Next-Generation  Sequencing

Now take a look at what is happening in genetics and genomics due to Next-Generation Sequencing: entire new fields are opening up, from targeted cancer therapies guided by precision diagnostics (something I work with every day in my regular work), to replacement of invasive amniocentesis by non-invasive pre-natal screening, to examination of the world’s built-up spaces by the microbial inhabitants that occupy them, to bringing to the International Space Station (via a SpaceX Dragon supply rocket payload) a DNA sequencer to look at potential contamination sources of their recycled water in microgravity, among other applications.

And what about the comparison of the Pony Express to modern jet travel? The comparison is not exactly equivalent (speak with anyone who uses NGS about the reliability of the vendor-assigned Qv scores for quality and how they compare with the established Sanger sequencing method by capillary electrophoresis’ Phred quality score and be prepared for an earful). Doing the comparison anyway, a state-of-the-art Applied Biosystems 3500xl can handle, depending on the configuration of POP-7 or POP-6 polymer and sequencing chemistry, some 1,104 samples at 600bp readlengths apiece, for a daily run throughput of 662 kilobases of QV20 CRL (Thermo Fisher’s PDF spec sheet here for clarification) sequence data.

Taking a mid-range laboratory that can afford a $250K instrument like the Illumina NextSeq (the Sanger instrument referred to above, if memory serves correctly, will be in the high 5-figures to low 6-figures), that throughput is now 129 gigabases in a two-day workflow. So we go from 662 kilobases to 65 gigabases per day, or from 662 kilobases to 65,000,000 kilobases. This is a factor of 98,000-fold. To be fair, the Sanger instrument(s) are optimized for ease-of-use, flexibility, and above all ‘gold standard’ sequencing quality.

So for those keeping score: Pony Express to the Jet Age, 59-fold in 100 years; Missile Command on the Atari 2600 to Rocket League on the XBox One, 13,000-fold in 40 years; an ABi 3730xl to an Illumina NextSeq 500, 98,000-fold in 12 years.

Electronic detection of DNA

In my daily work I come across some interesting questions, and one posed to me was about the intersection of microelectronics as a business and applied life sciences. (I use the term ‘applied life sciences’ as it may relate to protein, nucleic acid, or other chemical or metabolic analytes.) What follows is a bit of an abbreviated history of how DNA can be detected electronically.

Nanogen started early in the mid-90’s, and commercialized a $150K system around the year 2000. The Advanced Technology Center at the NCI purchased one for SNP detection, and this article says it wasn’t the number of SNPs detected or the sample throughput that differentiated it from Sequenom’s mass-spectroscopy based method, but rather accuracy. The fortunes of Nanogen, which rose quickly in the heady stock market days running up to the completion of the Human Genome Project, quickly dissipated in the following years. Friends of mine who where there selling the first systems to customers tell me that it just ‘didn’t work’; Nanogen continued in the diagnostics space until 2009, when it declared bankruptcy.

Clinical MicroSensors (called ‘CMS’ back in the day) was a company I came across during a job hunt in the late ’90’s when I lived in Southern California. The radio, component and phone company Motorola (for those who don’t remember, they had a popular phone called the StarTac once upon a time) purchased CMS in 2000 for some $300 million dollars. Let that sink in – $300M for a technology that combined DNA with electronic detection via a microchip. The technology turned into a company called Osmetech, received FDA clearance in 2006 for Cystic Fibrosis testing, and changed its name and went public as GenMark Diagnostics in 2010.

Its approach is a simple one, using PCR to replace the DNA backbone with chemical moieties to enable DNA to work as a conductor of electrical current, and by capturing the amplified product on the electronic chip with non-conducting DNA probes, the signal can be detected that way. Looking at GenMark’s application offerings, they focus on infectious disease (and CF as previously mentioned, as well as Warfarin sensitivity testing); looking at their financials for the past three years of steady losses, what I suspected is true: there are many alternative methods for detecting a PCR product.

A third company that started around the year 2000 was Combimatrix. They were among the first (later followed by NimbleGen) to use existing microelectronic fabrication facilities (individual memory location of RAM memory) as an individually-addressable ‘pad’ upon which DNA (or theoretically explored in their earlier days) protein could by synthesized. Today Combimatrix offers clinical testing of many types; the main feature of their technology was its ability to customize microarrays.

Another company to take a similar approach a few years later was febit, a Heidelberg-based microarray company whose instrument offered custom expression microarrays at the customer’s location. And going back to the early 2000’s, when Illumina began with the BeadChip technology, it was using photomicrolithography for making wells in silicon substrate, and then populating those wells with DNA oligonucleotide-coated beads.

Ion Torrent, Oxford Nanopore and Genia

I’d be remiss not to mention Thermo Fischer Scientific’s Ion Torrent technology, detecting ionic charge changes via CMOS technology and an ion-sensitive, proprietary layer. But I assume the readers here will be familiar with that already. Also Oxford Nanopore, and Roche’s upcoming Genia, are single-molecule technologies that also leverage microelectronics as a critical part of their DNA sequencing methods.

Today there is an entire journal devoted to the Lab on a Chip concept since 2001, and certainly innovation for miniaturization and to use existing, relatively inexpensive manufacturing techniques for making microelectronics are key drivers.

What’s next for microfabrication in the electronics industry to help life science? It could be solid-state nanopores, and Two Pore Guy’s may be only the beginning. At AGBT Hitachi showed proof-of-principle of what they are developing, at the 2 to 3 nm pore level, and using a piezo-electric substrate to slow down the speed of DNA strands being read across the junction. It may seem on the fringe of science fiction as several attendees I spoke with afterward expressed doubt about the commercial prospects for such complexity, but is certainly something to watch in the years to come.