Element Biosciences AVITI™ Sequencing System

A new approach to next-generation sequencing, a little-known company unwraps an instrument that competes directly with the Illumina NextSeq 2000 with a third more of the throughput at half the consumables price

For a technology that started in 2005 with the first 454 / Roche FLX next-generation sequencing (NGS) platform, followed in 2007 with the Solexa / Illumina Genome Analyzer, and then in 2010 with the Ion Torrent / Life Technologies / Thermo Fisher Scientific Personal Genome Machine (now known as Genexus), the NGS market has grown to be about a $6 Billion market.

Single-molecule technologies (aka ‘Third Generation’) systems have launched, with Pacific Biosciences Sequel II and Oxford Nanopore GridION and PromethION  leading the way, having introducing their first systems around 2012 and 2014 respectively.

In terms of overall market share, Illumina has about 70%, Thermo Fisher Scientific (TMO) at about 10%, Pacific Biosciences at 2%, Oxford Nanopore in that same range. (These numbers are both instruments and reagents for the overall market by revenues; thus the remaining 20% of the market is split among reagents-only companies such as New England Biolabs (NEB) or QIAGEN who sell into the NGS market. One other company that has an NGS instrument is MGI (formerly known as BGI), and has been locked out of the US market due to an injunction over intellectual property. MGI is expected to have freedom to sell their instrumentation in the US later in 2022.

For a $6B market, Illumina’s high market share has been problematic: first with their cancellation of their $1.2B acquisition of Pacific Biosciences (announced in late 2018, cancelled in early 2020) and now with an EU ruling pending about Illumina’s $8B acquisition of GRAIL (announced in fall of 2020). Another challenge to Illumina has been market saturation with their instru mentation: their revenues attributed to instrumentation has been relatively flat for the past 5 years (at around $600M to $700M per year). Of course, Illumina has updated their instruments regularly, with new iterations and models – the NextSeq line was originally called the NextSeq 500, then the NextSeq 550, after that the NextSeq 1000, and most recently the NextSeq 2000. Yet with all the proliferation of systems and capabilities, the instrument overall revenues are static.

Over the past few months a few friends have told me hints and snippets about Element Biosciences, that they were working on a new sequencer and that it would be worth keeping an eye on them. In March 2022, Element held a 3 hour launch event, which may (or may not) be available at a future date.

How Element Biosciences gets to a third more the throughput at half the price

The reason for the name Element Biosciences is to get to the fundamental ‘elements’ of next-generation sequencing, and re-invent it. In any creative endeavor, it pays (rather handsomely) to get to first principles: coined by Aristotle, used by Johannes Gutenberg and Elon Musk, getting to foundational and fundamental truths is a process of simplifying and reducing to get to absolute basics. In the realm of NGS, Element Biosciences (with their collection of some 15 patents) has reinvented the flowcell substrate (reducing background noise, in this case background fluorescence to increase contrast between the signal and the background), has reinvented the sequencing-by-synthesis concept (more about this later), and reinventing the instrument setup with not only independent flowcells but independent densities. There are plenty of other nuances involved (such as circular library molecules, also three distinct enzymes developed by Element with highly specific functions in their sequencing paradigm) but here will cover these three ‘elementary’ developments.

Regarding the substrate, we are entering a new world of new materials. For example, a few years ago Cardea Biosciences launched a biosensor based upon single-molecule-thick graphene. Another nanotechnology-enabled material called vantablack, composed of vertically aligned carbon nanotube arrays, that absorb 99.96% of all the photons that hit the surface. Element went through a lot of time and experiments to come up with a substrate where the background fluorescence was significantly lower than what is on the market.

It is notable that the principals of Element Biosciences have had a long tenure in technology development at different NGS companies – starting with the CEO Molly He, who was at Pacific Biosciences for two years developing optimized enzymes, and then at Illumina for another almost 8 years doing similar work; co-founder and CSO Michael Previte was at Life Technologies for two years and then at Illumina for seven years working on single-molecule and ensemble sequencing approaches; SVP of Engineering Francisco Garcia was at Illumina for 19 years from the very early days there. (N.B. – I worked closely with Francisco when I was a BeadArray Product Manager in 2003 to 2005 at Illumina, he’s a really great person to work with.)

Thus with expertise like this, with deep knowledge of the knock-on consequences of limitations encountered because of fundamental choices made early on, a huge amount of work and time went into optimization of this key foundation of their technology. In software engineering there is a term called “Technical Debt”, coined by one of the authors of the Agile Manifesto – basically you borrow time by prioritizing speedy delivery over perfect code; it is the idea of paying for the borrowed time ‘at a later time’ to refactor your code (i.e. clean it up). In this context, you build up technical debt by saving time and effort up-front by not spending additional weeks and months (or even a year) tweaking and refining and inventing in trying to get a new substrate with the right characteristics.

It appears that Element did the up-front investment, the equivalent of saving up to pay in cash, rather than taking out a loan, to come up with a flowcell substrate with extremely low background noise, in addition to being functionalized to have short oligonucleotides attached to the surface in order to build polymerase colonies (known as “polonies” per the historical name, developed in the George Church lab in the early 2000’s). That particular combination, of getting a surface material with very little background fluorescence, and to obtain an activated surface with biologically active oligonucleotides (or oligonucleotide analogues), is not an insignificant leap forward. The chemistry has then a lower bar to cross for signal to noise (or as Element likes to say contrast); the imaging equipment also has a lower bar to cross.

Element Biosciences’ sequencing by synthesis in two steps instead of one

With a substrate and a lawn of oligonucleotides, it is relatively easy to imagine a circular library molecule finding a universal sequence on the prepared flowcell. As a picture is worth a thousand words, well I grabbed some screenshots of a video that you can view here online (if you share your contact information that is). This illustrates a circular library molecule (specific Element kits are either integrated with partner manufacturers’ library or target enrichment kits, or as a stand-alone ‘convert your library to an Element library’ kit) that is amplified by what is known as rolling-circle amplification.

This polymerase colony (“polony”) is a cluster of library molecules, on the order of thousands of copies of the same single molecule that was circular and bound to a spot on the flowcell. Thus now there are thousands of identical clonal copies at that cluster.

Instead of binding a single labeled nucleotide at each of those thousands of locations in the cluster, AVITI technology uses a different approach: a single fluor with many octopus-like tentacle arms, each arm having a single nucleotide type (one of four, AGT or C) at the very end. Thus the detection is not a canonical 3’ to 5’ condensation reaction producing pyrophosphate; it is only entering a pocket of A=T, T=A, C=G, or G=C recognition of one of four nucleotide possibilities.

The ‘secret’ here is the strong binding; say there are 1000 molecules in the polony, each with a primer and an open ‘A’ base. The AVITI octopus-like labeled molecule is called an Avidite and has four different colors, and four different nucleotides attached. So Avidite-T will cross-connect to about eight individual copies of the 1000 molecules available with an exposed “A” base (remember A=T, T=A, G=C and C=G combinations). Thus 250 Avidite-T’s bind to the 1000 molecules in the polony, each Avidite-T binding to 8 individual target A bases. However no polymerase activity occurs; what happens is this binding is very stable in comparison to the other bases. Again, here’s a screencapture of the process – two different colors of Avidites are shown, one red (the “T” correct one) and one green-blue (a “G” incorrect one).

If an Avidite-G binds incorrectly to the exposed A base, it is transient and a weak binding. One of the 8 arms may be bound, but the other 7 do not stick. Thus mis-pairing of the bases increasing background noise is vastly reduced.

Now what may be confusing is the enzyme that locks in the A to T-Avidite molecule is called a Proprietary Polymerase, however it does not polymerize any DNA. Perhaps they should have called it a Proprietary Bindase, a Proprietary Recognitionase, or even perhaps a Proprietary Elementase. It just locks in the proper A=T-Avidite and selects for that, across 8 or so different DNA molecules. With around 250 of these arrangements per polony cluster (assuming 1000 copies per cluster) that’s a tremendous increase in signal.

Now with NGS as a sequencing-by-synthesis model, one problem is called phasing. This is where a small percentage of molecules do not extend one base, for one reason or another. And if that’s 99.9% efficient, that 0.1% grows exponentially on a per-cycle basis. Imagine our 1000 molecules in a cluster, and 0.1% do not extend; so after cycle 1 our 1000 molecules in phase now turn into 999 molecules. Thanks to geometric growth in the number of inactive molecules to give signal, phasing is a genuine problem with ensemble sequencing.

Element gets around this with a dark base with a reversible terminator, that is optimized for extending the growing nascent strand by one base on the original template.

Now going back to the Avidite labelled nucleotide with its ~8 tentacle arms, this reagent is another secret to how the sequencing can be so inexpensive: it has literally 1000-fold less concentration needed in the reaction, compared to Illumina’s dye-labelled, reversible-terminator chemistry. If you take a look at Illumina’s chemistry, these expensive labeled reversible terminators are visibly colored when you place them in the instrument, and the vast majority goes to waste as the working concentration is on the micromolar range. With Element’s Avidite reagents, they are used typically at 1/1000^th the concentration, in the nanomolar range. Thus 1/1000^th of the amount of expensive reagents are used per run.

A third innovation with Element’s AVITI instrument configuration

On top of the subtrate innovation, and the separation of SBS into two steps recognition then extension of a dark base, is the configuration of the instrument. It is not an insignificant task to design a new instrument with greater flexibility.

The QIAGEN GeneReader was going to innovate around having independent lanes with a proposed ability to run different libraries at different cycle numbers. The Ion Torrent technology was to lower the turnaround time to get sequencing data, in addition to a low cost of running an experiment at the lowest capacity at that time. Where Element has innovated is to allow two different flowcells to run independently of each other, on top of being able to control the density of the chip itself along with the read lengths.

There are digital counting experiments (including RNA-Seq, ChIP-Seq, and Olink Proteomics for quantifying up to 3000 circulating proteins) where only 35-base reads are needed. There are other experiments where 2×150 paired-end reads are not needed. There are still other experiments where a lower capacity is needed. Element offers one high-capacity flow-cell, and the capability to run shorter cycles, single-end reads, and lower density (the flowcells are not patterned, but rather are random).

High quality data and high volume of output

During the launch event, this diagram was showed by their CSO, and speaks to the high percentage of high quality bases in the sequencing output: >90% of all bases above Q30. With PCR-free protocols (that suppress errors introduced by PCR), >80% of all bases were above Q40 (!!).

Later on in the presentation, Shawn Levy (now SVP of Applications and Scientific Affairs, formerly an investigator at the Hudson Alpha Institute and group lead of their large NGS production sequencing facility) shows this impressive slide with per-base quality detail in a human reference sample, and then with PhiX control DNA.

As a side-note, I’ve known Shawn Levy for many years (going back to when he was at Vanderbilt), and for him to join Element Biosciences says a lot about his confidence in the company and technology. The fact that this company has developed all this technology and has launched a new instrument in the space of only 4 or 5 years is nothing short of remarkable.

A third more of the throughput at half the price

As a marketing person, this is how I summarize Element Biosciences’ new instrument the AVITI. They’ve obviously put in an enormous amount of work to integrate this new system easily into existing Illumina or other sequencing platform workflows, starting from converting libraries to making data analysis easier. See the following summary slides from the first part of the launch video.

Now for the best part: throughput and price. The quality of the sequencing is there, the ease of transitioning is there, what is the compelling reason to buy this over a NextSeq 2000?

Price and throughput.

The instrument at list price is $289K, some $46K less than Illumina’s NextSeq 2000. The AVITI flowcell provides 800M reads at 2x150bp for $1680, compared to $3800 (list price) for a P2 Illumina flowcell at 400M reads. The P3 Illumina flowcells cost $6000 (list price) and 1200M reads. On a cost/Gigabase basis, Element is around $7/Gb whereas the P2 is $31, and the higher-capacity P3 Illumina flowcell drives the cost/Gb down to $16. The times of these runs on both these manufacturers’ instruments are two days (48 hours).

So a straight-up comparison between AVITI versus the Illumina P3 on a per-run basis (not a per-flowcell is the AVITI has two flowcells):

AVITI: 1.6B reads, $3,360 cost / run, 480Gbases, $7/Gbase

Illumina P2: 400M reads, $3800 cost / run, 120Gbases, $31/Gbase

Illumina P3: 1200M reads, $6000 cost / run, 360Gbases, $16/Gbase

You can look at it this way: 33% more throughput at 46% of the cost.

Link to Element’s AVITI Product Page