Good day, ladies and gentlemen, and welcome to the fourth quarter 2019 Illumina earnings conference. At this time, all participants are in a listen-only mode. After the speaker's presentation, there will be a question and answer session. To ask a question during the session, you will need to press star one on your telephone. If you require any further assistance, please press star zero. As a reminder, this conference call may be recorded. I would now like to introduce your host for today's conference, Ms. Jacquie Ross, Illumina investor relations.
Good afternoon, everyone, welcome to our earnings call for the 2019 fourth quarter and full year. During the call today, we will review the financial results released after the close of the market and offer commentary on our commercial activity, after which we will host a question and answer session. If you have not had a chance to review the earnings release, it can be found in the investor relations section of our website at illumina.com. Participating for Illumina today will be Francis deSouza, President and Chief Executive Officer, and Sam Samad, Chief Financial Officer. Francis will provide a brief update on the state of our business, and Sam will review our financial results. This call is being recorded, and the audio portion will be archived in the investors section of our website.
It is our intent that all forward-looking statements regarding our financial results and commercial activity made during today's call will be protected under the Private Securities Litigation Reform Act of 1995. Forward-looking statements are subject to risks and uncertainties. Actual events or results may differ materially from those projected or discussed. All forward-looking statements are based upon current available information, and Illumina assumes no obligation to update these statements. To better understand the risks and uncertainties that could cause actual results to differ, we refer you to the documents that Illumina files with the Securities and Exchange Commission, including Illumina's most recent Forms 10-Q and 10-K. With that, I will now turn the call over to Francis.
Thank you, Jacquie. Good afternoon, everyone. Illumina had a solid end to 2019, with fourth quarter revenue of $953 million, up 10% from the fourth quarter of 2018. Highlights included 22% year-over-year sequencing consumable growth and stronger than expected IVD partnership and non-DTC array revenue. This more than offset softer sequencing system revenue and weaker than expected DTC revenue. While some variation of mix should always be expected in our dynamic industry, we are pleased to have delivered higher than expected revenue in the fourth quarter. For 2019, we delivered revenue of $3.5 billion, up 6%, and in line with the guidance we set in July of last year. We shipped more than 2,400 sequencing systems, the most in Illumina's history. We achieved our 2019 NovaSeq goal with approximately 320 shipments, slightly higher than 2018.
Total sequencing consumable revenue grew 14% and surpassed $2 billion for the first time, including more than $1 billion of high throughput sequencing consumables, and total data generated by Illumina sequencers increased 50%, highlighting the rapidly growing demand for genomic information. Back to the fourth quarter, NovaSeq consumable pull-through was at its highest level of the year, and indeed the highest since the platform was launched, driven in part by the U.K. Biobank, which is now operating at scale. We shipped more than 100 NovaSeq systems in the fourth quarter. As we expected, this was more than twice the number we shipped in the first quarter of 2019. NovaSeq pull-through per system for 2019 was approximately $1.2 million, and looking forward, we are targeting a pull-through range of $1.1 million-$1.2 million per NovaSeq system in 2020.
As we enter NovaSeq's fourth year, HiSeq consumables continue to decline as expected and were below $100 million in the fourth quarter for the first time since the NovaSeq launch. At the end of the year, we completed a review of our HiSeq customer list, indicating an active HiSeq install base of approximately 1,300. This review identified approximately 600 currently inactive HiSeq systems that had previously been included in our install base. Outside of this review, there were approximately 200 decommissions reported to Illumina in 2019. Moving to mid throughput, NextSeq delivered a record number of shipments in 2019 and now has a global install base of about 3,600. Fourth quarter NextSeq shipments were lower than expected due to customer timing. Demand for NextSeqDx continues to grow and represented almost a quarter of 2019 shipments, up from approximately 10% last year.
It was a record quarter for NextSeq sequencing consumable revenue, driven once again by oncology and NIPT. NextSeq pull-through per system improved from last quarter and was at the lower end of the $130,000-$160,000 target range. With the launch of our new NextSeq 1000 and NextSeq 2000 systems, we will update our target pull-through range when we have a sizable install base of the new systems. We expect to ship approximately 500 NextSeq 1000s and 2000s this year. While most NextSeq 550 prospects will transition to the new systems, we expect the NextSeq Dx pipeline to remain strong given the unique positioning of our regulatory approved system. We're looking forward to shipping the first NextSeq 2000s later this quarter and are very pleased to announce that we have already received our first orders.
Moving to low throughput, system revenue was below our expectations in the fourth quarter, primarily due to MiniSeq. Demand for MiSeqDx continues to exceed our expectations, with particular strength in China, where the system was cleared by the China NMPA in August of 2018. Earlier this month, the MiSeqDx was approved by the PMDA in Japan, which we expect to contribute to placements in 2020. Turning to low throughput consumables, it was a record revenue quarter. MiSeq consumable pull-through grew closer to our target range of $40,000-$45,000, and MiniSeq pull-through was at the low end of the $20,000-$25,000 range. Back to sequencing consumables. Total revenue of $2.1 billion grew 14%, or over $250 million in 2019. Just over 40% of our sequencing consumable shipments were for clinical, which includes testing for oncology, reproductive health, and genetic disease and other.
In total, clinical sequencing consumables grew about 20%, or approximately $130 million in 2019 to approximately $830 million. Oncology continues to represent about 20% of total sequencing consumables and grew faster than total clinical consumables in 2019 due to increased adoption of panels, including comprehensive genomic profiling. As more tests like FoundationOne CDx and Guardant360 receive coverage as companion diagnostics, demand for Illumina sequencing continues to grow. Additionally, clinical trials like Guardant's LUNAR drive increased sequencing consumable utilization in oncology testing. Reproductive health once again represented a little more than 10% of sequencing consumables, primarily reflecting continued growth in NIPT due to broader coverage in EMEA, where our VeriSeq NIPT solution had 80% sample volume growth and growing adoption in China. Reproductive health continues to grow in the U.S. at a more modest rate compared to EMEA and China.
Within clinical, almost 10% of our sequencing consumable revenue is related to genetic disease testing, which grew slightly below the clinical average. Growth is driven by companies like Centogene, which has built a genomic repository of over 450,000 patients from over 125 countries and is working to help diagnose patients' genetic disease and collaborating with pharma partners to find cures. It also includes a portion of revenue from genetic testing companies like Ambry, who offer tests tailored toward genetic disease diagnosis, among other clinical tests. Turning to research and applied, shipments of over $1.2 billion represented just under 60% of our sequencing consumable shipments. As a group, research grew over 10% in 2019, driven by genetic disease and cancer. Genetic disease research includes population genomics initiatives such as the U.K. Biobank and Million Veteran Program.
Cancer research was also a strong contributor to growth and includes projects like the Cloud Initiative at St. Jude Children's Research Hospital. The St. Jude team is building a database to access whole genomes of 10,000 pediatric patients and survivors, helping researchers to gain valuable insights into the genetic causes of pediatric cancer. Other research categories include cell and molecular biology research, microbiology, and infectious disease testing. This includes projects like J-GRID, the Japan Initiative for Global Research Network on Infectious Disease, which is a collaboration between nine countries in Asia and Africa and utilizes Illumina sequencing to research microbial diseases.
Moving to sequencing services and other, revenue of $124 million was up 19% from the same quarter a year ago, largely driven by upfront revenue from the Roche deal, partially offset by GeL, which declined as expected to almost zero in the fourth quarter of 2019, ahead of the clinical ramp-up later this year. Finally, arrays delivered revenue of $116 million, down 12% from the same quarter in 2018, due to continued headwinds from our direct-to-consumer customers, offset in part by array growth in genetic disease research. Before I hand the call over to Sam, I'd like to comment on the novel coronavirus outbreak. Our immediate focus has been our colleagues in China. Our thoughts are with the families and communities impacted. Over the last few weeks, Illumina has been engaged in a number of ways to help manage the coronavirus outbreak.
Scientists have already used Illumina sequencers to identify and publish the genomic profile of the coronavirus into the public databases, which is a critical first step to enable the development of diagnostic tests and ultimately potential vaccines. Our team is actively working with Chinese CDC labs to prepare coronavirus NGS testing protocols and provide the necessary training. We're also working with our supply chain team to ensure that systems and consumables are delivered to labs working with novel coronavirus as quickly as possible. We plan to share these NGS testing protocols with customers to support the global infectious disease community as it mobilizes to address this threat. Further, we're exploring philanthropic programs and collaborations to ensure novel coronavirus sequencing is available by providing sequencing and consumables to those who need it to fight this epidemic. With that, I'll hand the call over to Sam.
Thanks, Francis. As discussed, fourth quarter revenue grew 10% year-over-year to $953 million, driven by 14% growth in sequencing, offset by a 12% decline in microarray. Total sequencing revenue of $837 million grew 14% from the fourth quarter of 2018 and represented 88% of total revenue, compared to 85% in the same quarter last year. Sequencing consumable revenue of $572 million grew 22%, or over $100 million, compared to the fourth quarter of 2018, while sequencing system revenue was down slightly sequentially and down 12% compared to last year. With regards to sequencing systems, we are reviewing installed base counts to proactively identify the active systems. We have completed the work for HiSeq and are now reviewing NextSeq, MiSeq, and MiniSeq. We will revise those installed bases in the next quarter or two as we complete the analysis.
Importantly, this does not change the reported revenue in any way. We do expect system counts to decrease while the pull-through per system will therefore increase. Sequencing service and other revenue of $124 million was down $14 million sequentially due to lower IVD licensing and milestone revenue and lower GeL volumes in the fourth quarter, but up 19% year-over-year. Fourth quarter results included an upfront payment associated with the recently announced Roche partnership. Arrays represented 12% of total revenue in the fourth quarter, compared to 15% in the fourth quarter of 2018 and to 16% in the fourth quarter of 2017. Array systems were up $2 million sequentially, but down $5 million from a particularly strong fourth quarter of 2018.
Array consumables grew $18 million sequentially due to DTC seasonality, with one customer ramping ahead of the holiday season, but were modestly down from the same quarter last year. Array services were down $6 million sequentially and down $8 million, or 32%, year-over-year due to lower demand from our DTC customer. Moving to regional results, Americas revenue grew 5% versus the prior year quarter, with growth in sequencing consumables and IVD partnership revenue partially offset by DTC headwinds. EMEA delivered a record revenue quarter with 19% growth from the prior year quarter, including a strong contribution from the U.K. Biobank, which is now sequencing in full production mode. Greater China grew 21% from prior year quarter, with an easy year-over-year comp associated with tariff-related stocking in China in the fourth quarter of 2018.
The region grew 2% in 2019, with slower research offsetting very strong growth in clinical. Finally, APJ revenue of $73 million was up 4% from the fourth quarter of 2018, driven by genetic disease research and microbiology driving sequencing consumable growth both sequentially and year-over-year. For the full year, the region grew 6%. Moving to gross margin and operating expenses, I will highlight non-GAAP results that include stock-based compensation. I encourage you to review the GAAP reconciliation of these non-GAAP measures, which can be found in today's release and the supplementary data available on our website. Please note that all subsequent references to net income and earnings per share refer to the results attributable to Illumina shareholders.
Non-GAAP gross margin of 70.2% was roughly in line with expectations and decreased approximately 230 basis points compared to the third quarter, with lower IVD licensing and development revenue, in addition to variances in production and lab service absorption, partially offset by product mix. Year-over-year, fourth quarter non-GAAP gross margin increased over 100 basis points, primarily due to product mix and higher IVD licensing and development revenue, partially offset by lower DTC service volumes. Non-GAAP operating expenses of $372 million were up $42 million from the third quarter of 2019, largely reflecting the timing of OpEx spend weighted towards the end of the year and were better than expected. Non-GAAP operating margin was therefore 31%, down from 36.1% last quarter. The non-GAAP tax rate of 18.5% was up from last quarter and higher than expected due to income mix in various tax jurisdictions.
For the fourth quarter of 2019, GAAP net income was $239 million, or $1.61 per diluted share, and non-GAAP net income was $252 million, or $1.70 per diluted share. Moving to cash flow and balance sheet items, cash flow from operations was $443 million, helped in part by a $58 million sequential decline in inventory. This is part of an initiative led by our operations and supply chain organizations to optimize our working capital. DSO of 55 days compared to 54 days last quarter. Fourth quarter capital expenditures were $57 million and free cash flow was $386 million. We repurchased $63 million of stock in the fourth quarter, leaving $226 million available for share repurchases under our current plan. We ended the year with approximately $3.4 billion in cash equivalents, and short-term investments. Our weighted average diluted share count for the quarter was 148 million.
Moving to guidance, we expect full year 2020 revenue to grow in the range of 9%-11%, or $3.86 billion-$3.93 billion, representing an increase of approximately $354 million at the midpoint. Given the new system launch and ongoing weakness in DTC, we are expecting revenue linearity to be similar to 2017, which suggests just below 22% of revenue in Q1, approximately 24% in Q2, approximately 26% in Q3, and approximately 28% in Q4. For the full year 2020, and at the midpoint of our revenue guidance range, we expect sequencing revenue to grow approximately 14%. This includes sequencing consumable growth around 17%, and we expect sequencing system revenue to grow year-over-year, reflecting our NextSeq 2000 and 1000 launch more than offsetting the expected step down in NovaSeq shipments. We also expect sequencing service and other to be roughly flat year-over-year.
We expect array revenue to be down approximately 15%, reflecting ongoing weakness in DTC. Note that DTC revenue represented approximately 50% of total array revenue in 2018 and decreased to about 40% in 2019. For 2020, we expect DTC to represent approximately 30%. From a regional perspective, we expect China to grow in the high teens, driven by clinical sequencing. We expect full year non-GAAP gross margin to be roughly in line with 2019. We expect operating margin to be approximately 30%, and we expect the 2020 tax rate to be higher than full year 2019 due to a number of one-time discrete tax benefits in 2019 that are not expected to repeat in 2020. We therefore expect GAAP earnings per share in the range of $6.45-$6.65, and non-GAAP earnings per share in the range of $6.80-$7.
We expect diluted shares outstanding in 2020 to be about flat compared to Q4 2019. Note that following the termination of the Pacific Biosciences merger agreement earlier this month, Illumina paid a $98 million reverse termination fee to PacBio. Additionally, Illumina will pay $34 million in continuation advances that may be repayable to Illumina if PacBio enters a change of control agreement or raises at least $100 million within a given time frame. The impact of these payments is not reflected in our EPS guidance, pending the valuation of these amounts this quarter. Moving to the first quarter of 2020, we expect total revenue to be between $850 million and $855 million as customers consider and adjust their plans following our NextSeq 2000 launch.
Non-GAAP operating expenses are expected to increase approximately 600 basis points as a percentage of revenue on a sequential basis, primarily due to our bonus accrual reset at the start of the year. We expect the first quarter non-GAAP EPS to be between $1.20 and $1.25, and GAAP EPS to be $1.11- $1.16. With that, I'll hand the call back over to Francis.
Thank you, Sam. We're off to a strong start in 2020. The UK Biobank is sequencing at scale. We are contracted with GeL to provide sequencing services starting in the middle of the year, significantly strengthening our PopGen visibility relative to where we were a year ago. We launched our most innovative system to date with the NextSeq 1000 and 2000. The system offers the highest cluster density flow cell of any NGS system, driving down the cost per gigabase for mid throughput users. We succeeded in our ambitious target to fully integrate the hardware accelerated best in class pipeline that we acquired with Edico just 18 months ago. We're excited about our TruSight software suite, a potentially transformative solution that simplifies genetic disease diagnosis and reduces barriers to adoption. I look forward to sharing updates on it after launch.
We continue to extend our clinical portfolio and both TSO 500 and TruSight NIPT are progressing through regulatory. We're expanding our capabilities through partnerships with the world's leading clinical companies, including Qiagen, Roche and Adaptive, to deliver the most compelling IVD menu available on our clinical grade sequencing systems. As our customers discover more biological insights by sequencing at greater depths and volumes across new and emerging applications, the growing clinical utility of genomic information is becoming increasingly clear. This, coupled with growing community awareness and physician adoption, will enable more patients to benefit from the promise of genomics. With that, I'll invite the operator to start the Q&A.
Thank you. As a reminder, to ask a question, you will need to press star one on your telephone. To withdraw your question, press the pound or hash key. Please stand by while we compile the Q&A roster. Your first question comes from the line of Tycho Peterson with JP Morgan. Your line is open.
Hey, thanks. I'll try to ask just a couple quick ones up front. On the fourth quarter, can you quantify the Roche milestone? It looks like the partnership fee stepped up $15 million-$20 million. Was that all Roche in the fourth quarter? Looking ahead on instruments, coming off Q4 where you were light on instrument revenues, are you doing anything different in terms of NovaSeq pricing to try to catalyze the remainder of the HiSeq upgrade cycle? It looked like NovaSeq ASPs did decline a bit in the fourth quarter. That's the first question. On NextSeq, just curious, as you think about that, is it an upgrade cycle to the installed base or market expansive as those two new systems get rolled out into the market? Thanks.
Yeah, thanks, Tycho. This is Sam. On the Roche milestone, we didn't quantify that. It did come in better than expected. The sequencing and other line came in better than expected in Q4, I would say approximately $20 million or so, better than expected in terms of sequencing and other. We didn't quantify exactly how much of that was Roche. You can think about it as IVD revenues that we had in Q4, that the majority of which was driven by Roche. With regards to your second question, I'll take that one, and then Francis can comment on NextSeq. With regards to your second question around the ASP on NovaSeq, it's not that we're doing anything special around reducing the ASP or reducing the price or giving any discounts to customers. This is natural and driven by multi-unit orders that we have across 2019 as well.
You can look at it as the UK Biobank, a number of placements that we had with the UK Biobank that usually have a lower ASP. We've had some NovaSeq placements in emerging markets, which also have a lower ASP. It's natural dynamics there when you place multi-unit orders with some of our customers, but nothing specific on lowering the price across the board.
Thanks, Sam. Hi, Tycho. In terms of the NextSeq 1000 and 2000 target base that we're going after, we're really focusing, as you pointed out, this to be an upgrade cycle for customers off the NextSeq 550 today. That's where we expect the bulk of the demand to come from for the NextSeq 2000 that launches this quarter and then the 1000 that launches later in the year. We also expect a smaller bolus of orders to come from some MiSeq customers. We highlighted a couple of quarters ago that we're starting to see MiSeq customers that are migrating up and buying bigger instruments. We expect those customers to be looking at the NextSeq 1000 and 2000. That's where we expect the bulk of the customers to come from.
There is a chance that there might be a small number of HiSeq customers that look at the NextSeq 2000, but that's likely to be a very small number.
Your next question comes from the line of Dan Brennan with UBS. Your line is open.
Great. Thanks for taking the question. I just wanted a little more color on kind of NovaSeq. I know you're talking about placements being down in 2020. Can you give us some help on how much of the install base you expect to kind of have upgraded by the end of the year and kind of any more color on the level of it being down? Secondly, just related to HiSeq consumables, since you are kind of giving more granularity there, what's kind of baked in within your consumable guidance for HiSeq consumables in 2020?
All right. If we look at the NovaSeq upgrade cycle, here's how the numbers are playing out. We believe that under half of the HiSeq X customer base has started the NovaSeq transition, we still have over half of that install base to go. That's going to play out between now and 2024, when we end of life the HiSeq. That's the numbers we expect to happen. In terms of the number of NovaSeq, last year was a very strong year in terms of NovaSeq placements, which is something we're really excited about, especially given that it's the third year of the NovaSeq upgrade cycle. We expect that number to come down this year quite naturally, we expect that to play out similarly over the next couple of years.
With regards to HiSeq, Dan, we're not giving specific information on the level of consumables or what to expect, even in terms of pull-through. We've backed away from giving pull-through ranges on HiSeq and total consumables. What we've mentioned for Q4, as you heard on the prepared remarks, was that consumables for HiSeq were less than $100 million in the quarter. They continue to come down. Obviously, the expectation, I think intuitively as you would expect in 2020, that HiSeq consumables will continue to come down. We haven't given specifics on what that dollar number is.
Your next question comes from the line of Steve Beuchaw with Wolfe Research. Your line is open.
Hi, good afternoon. Thanks for the time here. I want to ask one on China and one on the NextSeq launch. As it relates to China, I appreciate that you've given a view dating back to a few weeks ago that there's an expectation for some recovery of growth in China. I wonder if you could speak to the balance of what you're seeing there between funding dynamics and as opposed to maybe just easier comps, and then of course, the broader questions that we have now around the environment, given the viral outbreak, to what extent that is or is not a factor that one considers. Secondarily, now that we're a few weeks into, and I appreciate that's still early, the NextSeq launch. How are your higher throughput customers, who might also be NovaSeq buyers or owners, how are they thinking about using this system?
Is it fair to say that we're comfortable with the trajectory on NovaSeq orders being largely unimpacted by the NextSeq, given as pretty attractive price per G dynamics on the P3 flow cell? Thanks.
Yeah. Thanks, Steve. I'll start with the China question, and then I'll get to the NextSeq question. In terms of China and what we're looking for, we saw good growth from China in Q4. A little of it was based on a friendlier on-year comp. We are seeing growth, and we knew that would be lumpy. In terms of this year, we are expecting, as we said, growth in China. That's going to be primarily driven from the clinical markets. More specifically, oncology continues to be a very strong area for us in China, and we expect that to continue to play out this year.
In terms of the coronavirus, I'll say first and foremost, our thoughts are with the families and communities that are impacted by the virus, and one of our top priority is to make sure we're doing everything we can to help and making sure that our teams stay safe. In terms of the financial impact of that, it's too early to call the financial impact either in Q1 or for the year. As we get more clarity, as this plays out, we'll certainly communicate transparently with investors. As we thought through the impact, it's possible that, if the epidemic grows, and if patients start to avoid going to hospitals, there could be an impact to NIPT and cancer testing in China, but it's way too early to call that impact. We're going to continue to watch that situation, and again, we'll communicate with you what we see.
We are deeply engaged in working on the coronavirus. We have teams working on that around the world with the Chinese CDC teams as well as CDC teams around the world. In terms of the NextSeq launch and how are our high-throughput customers, especially some of the legacy HiSeq customers, thinking about it. As I said, we really expect the NextSeq 1000 and 2000 to appeal to existing NextSeq customers. That's going to be the bolus of the upgrade cycle. There's going to be some interest from MiSeq customers that are upgrading and a little interest from existing high-throughput customers, but not much. The reason for that is if you look at the output of the NovaSeq and you look at the output of the NextSeq, what we've done is we've tried to cover the ground, but there's very little overlap.
At the very high end, if you go with the high output flow cell of the NextSeq 2000, you can get to an output of 300 gigabases. At the low end of the NovaSeq, with the Prime flow cell, you can get 500 gigabases of output. There's a clear gap there, and so you self-identify. Most of our high-throughput customers, we think will go with the smaller flow cell on the NovaSeq, and the NextSeq 2000 and 1000 will continue to appeal to that mid-throughput customer base.
Your next question comes from the line of Bill Quirk with Piper Sandler. Your line is open.
Great. Thanks, and good afternoon, everybody. Quickly, thinking about the clinical oncology growth over the next couple of years, there's several very innovative non-invasive screening options as well as recurrence monitoring tests coming online. How should we think about this business overall trending relative to 2019? Could it possibly inflect back to the, I think it was 30%, maybe a year or two ago? Separately on coronavirus, are there any analogous situations to other outbreaks, things like MERS, for example? If so, have you studied the impact on what happened, say, in the Middle East region to your business during that outbreak and taken a look at that relative to coronavirus? Thanks, guys.
Great. Well, thank you, Bill. I'll first talk about the clinical oncology market, specifically touching on the potential impacts of the screening opportunity and emerging monitoring opportunities on the growth rate, and then I will talk about the coronavirus. In terms of clinical oncology testing, today, the bulk of the revenue that we get from clinical oncology, and the bulk of the growth, frankly, is coming from customers that are using NGS-based testing, various types of panels, for therapy selection. That's sort of the bulk of the revenue that we're seeing in clinical oncology. That's the most penetrated part of the clinical oncology market. Frankly, it's still at its early stages. It's still about 8% penetrated. As we look at the other areas in clinical oncology, we see really exciting opportunities and potentially much larger opportunities coming from monitoring and screening.
Both of them are in the very, very, very nascent stages. Customers like Guardant and GRAIL and Freenome are doing some really exciting work there. The monitoring opportunity is starting to play out maybe a little bit further along than the screening opportunity today, where we're seeing the early use of liquid biopsies to monitor the effectiveness of a therapy, to look for MRD and for a recurrence of a cancer. There, customers like Guardant are starting to sell tests into that space. Again, very early stage. We're starting to see revenue from it, and we think this bit represents a potential large growth opportunity for us in the future years. Screening is a little bit earlier, again, super nascent, but hugely exciting.
I don't know if you've had a chance to look at the results that companies like Freenome and GRAIL have been publishing over the last few months at ASCO and ESMO, but the GRAIL results, for example, show that they are able to, in those studies, have very low false positive rates. Their numbers indicate maybe less than 0.5% false positive rates and high levels of specificity for over 20 different types of cancer. They're still in the study stage. They're talking about having a test, maybe end of this year, sometime next year. Obviously, that represents a very meaningful market opportunity in terms of impact on human life and represents a very big opportunity in terms of revenue for those companies and also for Illumina. That's something that'll play out not much this year, but in future years.
If you think about coronavirus and how that's playing out, we looked at how MERS played out and SARS played out, it's a little bit different because the market's a little bit further ahead in terms of the state-of-the-art NGS testing. The way this will play out, we think, is that NGS will play a number of critical roles in combating this coronavirus outbreak. While we believe first-line testing of patients will be done with RT-PCR, NGS will be used in a number of specific ways. One, we've already seen it play out, where NGS will be used to initially sequence the virus and create the publicly available reference needed to help develop diagnostics and ultimately cures. We started to see that play out already with coronavirus, we saw that previously with SARS. NGS will also be used to confirm the strain in patients.
It'll be used for detecting novel mutations in the viral genome. We saw that in other outbreaks, not just MERS and SARS, but we saw that with Ebola. Also it'll be used for epidemiological surveillance, which will be very important in terms of tracking this outbreak. Finally, we expect NGS to also be used as last resort testing for patients with inconclusive tests. For example, when a patient presents with suspicious symptoms, but the RT-PCR tests are negative for coronaviruses and also for other viruses with similar symptoms like influenza, then we expect an NGS test to be used. That's our point of view based on what we've seen with previous outbreaks.
As I said earlier, our teams around the world are currently engaged and have been working, for example, in China with the CDC team since December, and we're working with CDC teams around the world now. We're also engaged with philanthropic organizations and collaborations to make sure that NGS testing is available globally to monitor this outbreak.
Bill, maybe to emphasize a couple of data points on the clinical business and oncology as well, back to your initial question. For 2019, per our comments earlier, our sequencing consumables in terms of clinical grew by approximately 20%, and in terms of oncology, sequencing consumables grew above that, so over 20%. When we look at also our top customers in oncology, they grew also above the oncology average. Just some data points around oncology and clinical growth.
Your next question comes from the line of Derik de Bruin with Bank of America. Your line is open.
Hi, good afternoon. Hey, a couple of questions. The first one, can you talk a little bit more about the HiSeq decommissions? I'm sort of surprised by the number and just sort of talk about the labs that are not using it. Are those customers now outsourcing more? Did they buy NextSeq or something else? That sort of leads into the next question, which is, surprised to see that the NovaSeq consumable guide for 2020 was about $1.1 million-$1.2 million per box. I'm just wondering, if your HiSeq customers and those instruments are coming down, I'm just surprised why that NovaSeq number isn't higher than it is.
All right. Thank you, Derik. That's the result of a focus we are putting on getting closer to our customers, getting more visibility into how our customers are using the instruments. We spent some time really looking through the instruments to see which instruments are active, which instruments are not active. That's a new model for us, understanding what instruments are being used out in the field. The total number, 800, isn't what got decommissioned last year. That's about 200. The total number is sort of a catch up to say, okay, of all the instruments we have out in the field, how are they being used? Which ones are being used a lot? Which ones are being used less? What we came down to is about 800 that we feel are largely inactive.
It doesn't change anything in terms of the revenue we're getting from HiSeq or consumables. What it does mean is that there are fewer active instruments in the field, and the ones that our customers are using actually are being utilized more than the previous model had estimated, and that the pull-through per instrument at those labs is actually higher than the old model had estimated. Those customers aren't doing anything different. They aren't outsourcing more. They still have both the active and inactive instruments, but when we looked at it, they were using some instruments more. That's the dynamic of how it plays out.
In terms of your second question, Derik, with regards to NovaSeq, yes, our guide for 2020 is $1.1 million-$1.2 million pull-through per instrument. We're very pleased in 2019 with the performance that we've seen in terms of the pull-through on NovaSeq. As we mentioned, it was $1.2 million for the year, it was a record quarter in Q4. Our expectation now in the fourth year of launch in 2020, is that as we start seeing some of those lower throughput customers convert and move over, and even newer customers that are using NovaSeq with the S Prime and the S1 flow cell, that you will start to see lower throughput from those customers impacting the average. That's natural in the fourth year of launch for the instrument, that's the driver for the $1.1 million-$1.2 million pull-through guide.
In addition to, as Sam pointed out, some of the smaller customers we expect to come online this year, and part of the reason for the guide we gave, there will be an offsetting influence happening as you see some of these large PopGen customers continue to sequence, like the UK Biobank, that's really running in full force, as well as NHS Commissioning coming online, and so that'll be the offsetting factor that'll play out this year.
Your next question comes from the line of Dan Arias with Stifel. Your line is open.
Afternoon, guys. Thanks for the question. Apologies for a little bit of a cold here, Francis, on the NextSeq, can you just talk to the rationale for the launch of the 1,000 unit in the context of just thinking back to the NovaSeq 6000 and 5,000? Back then, it seemed like customers that upgraded really wanted the higher capacity option and to not have to wait around to acquire it. I guess I'm curious about the way in which you see this being different. Maybe relatedly, anything more specific you can say to the P3 flow cell launch later this year in terms of timing? I'm curious about how much we should expect demand for that configuration to be captured in 2020. Thanks.
Sure, and I hope your cold gets better soon.
Thanks.
I will answer the NextSeq question first. Why do the 1000 if we expect the majority of customers to buy the 2000, and then I'll talk about P3. The strategy of having the higher and lower-end instruments is one that we've used many times. You pointed out with the NovaSeq 5000, 6000, we have the HiSeq 3000, 4000, we have the HiSeq 2000, 2500. What this allows our sales teams to do is it allows us to start the conversation with customers at a lower price point. Some customers do buy the lower instrument, but as you pointed out, what happens in that conversation is the vast majority of cases, they end up buying the higher priced instrument. That's played out in every one of those instances I talked about going back a decade, right?
It's important for our reps to be able to have that lower priced instrument so they can start the conversation, and that's been the strategy. In terms of the P3 flow cell, the way we expect this to play out is that the customers for the first couple of full quarters of NextSeq 2000 are going to be excited about the fact that they can buy a flow cell whose output is comparable to today's NextSeq. They can compare the new machine on an equivalent output using equivalent workflows from the old machine to the new machine. We do expect that a lot of our NextSeq customers are going to buy the instrument with an eye to their future business and will be very attracted by the higher output and the economics associated with the P3.
In terms of timing, we said we expect that to come towards the end of the year. Obviously, though, it's public now and the specs of that flow cell are public now, so it's something that customers know about and can plan for.
Your next question comes from the line of Doug Schenkel with Cowen. Your line is open.
Hey, good afternoon. Just a couple of things on NovaSeq. First, just looking back, regarding guidance for consumable pull-through per box of $1.1 million-$1.2 million per Nova, how does that compare to 2019? I guess, is the reduction in HiSeq consumable revenue below $100 million, which is a little bit below where I think you guys have been tracking the last few quarters. I'm just wondering if that's indicative of larger programs on HiSeq X's being largely completed at this point, and some of those large customers with those projects now increasingly moving over to Nova. Those are the looking back topics I'd like to cover.
Looking ahead, is the assumption that NovaSeq placements are lower in 2020 versus 2019 a function of trying to hedge for the potential that there's some Nova cannibalization, a function of reducing the number of HiSeqs that are out there to replace, so maybe having a little bit more visibility on what the replacement outlook is, a function of being in year four of the rollout, or maybe just a little bit of all of the above? Finally, is there anything structural that would prevent you from implementing the 2-color chemistry on a higher-end platform like NovaSeq or actually on a NovaSeq in the future? Thank you.
Hey, thanks, Doug, for the question. I'll start with the first one, and I'll transition to Francis for the second and third. With regards to NovaSeq, so for the full year 2019, we communicated that pull-through was $1.2 million per instrument, and going forward, we expect it to be, in 2020
$1.1 million-$1.2 million. Essentially, the high end of the range is equivalent to what we saw in 2019. With regards to your question on HiSeq, I think it's exactly right. As we get deeper into this upgrade cycle, as we get deeper into this cycle where our customers are transitioning from HiSeq to NovaSeq, as some of this work transitions over and completes on HiSeq and moves over to the new instrument, you do expect to see consumables come down on HiSeq, which is what we've indicated, is that we expect in 2020 to continue to see HiSeq consumables come down and obviously, NovaSeq consumables to increase.
Doug, you asked about a couple of other things. One, in terms of NovaSeq placements, why are we expecting that to be lower this year given that we've had two back-to-back very strong years of NovaSeq placements?
Is it being influenced, you asked, by NextSeq and cannibalization there? You asked, is there a structural reason or any structural reason why the blue-green chemistry couldn't, in the future, be implemented on NovaSeq? In terms of NovaSeq placements, as we looked at this year, the work we see to be done this year in terms of the NovaSeq upgrade cycle is to really go after the smaller core labs, right? That's the upgrade cycle that got activated last year with S Prime, with S1s, and then S2's new pricing. As we look at the pipeline this year, we expect it to be a lot more full of the 1z, 2Zs with the small core labs. That's a dynamic we expect to be playing out in our customer base and in our sales teams.
That will result in a total number that's smaller than some of the big, multi-unit purchases you saw last year in Q3, Q4 with customers like the UK Biobank, for example. Instead of doing 10-unit placements, we expect more deals to be one to two units. That's frankly the biggest driver in terms of the placement number being down this year compared to last year. We don't expect much cannibalization between the NextSeq and the NovaSeq. They target very different price points, both in terms of capital but also in terms of cost per G. There's no overlap in terms of the output. When you add those three variables together, it's very easy for customers to self-select. If you have the sample volume, to allow you to go purchase a NovaSeq, you will always do that because you get superior economics.
Even with the new NextSeq 2000 running the P3 flow cell, you still get superior economics on the NovaSeq if you have the sample volume. If you don't have the sample volume, it'll never cost in for you to go to a NovaSeq. There's very little overlap in terms of the segment they're targeting. There may be some, but we don't expect much. Then your question about structural reason, why we can't use the blue-green chemistry, NovaSeq, but frankly anywhere. The answer is there isn't any. We view that as a core architectural component in our toolkit now, as we do other elements of things you saw in NovaSeq.
For example, the hardware acceleration toolkit that we built into NextSeq 1000 and 2000 based on the Edico technology, that's also another core architectural component, and there's no reason why that can't show up in any of our future instruments, both up and down the portfolio. What'll decide whether it does show up, frankly, is the design point for the next instrument we put out and if we have something better in the toolkit, and that's always sort of an active debate internally. At any given point, we'll pick the best components we have, the design point we're going, but there's absolutely no structural reason, as you point out, instrument or a lower throughput instrument.
Your next question comes from the line of Puneet Souda with SVB Leerink. Your line is open.
Yeah. Hi, Francis. Thanks. My first question is, I was hoping to get a better view into the NHS clinical ramp, and the level of visibility you have there. You made comments around clinical samples ramping in second half this year. This is one of the largest project, if not the largest. And I appreciate that you're expecting 300,000-500,000 patients sequenced by 2025 target, but how do we bridge that to the 5 million potential genomes that NHS had laid out earlier? And what should we expect in the second half this year? On NextSeq, I completely get your comments around 2000 uptake. Our checks are suggesting the same among academic customers, smaller labs.
I just wanted to understand, in terms of those that have diagnostic instruments or instruments in diagnostic settings, NextSeqs, and MiSeq, what's your expectation for those customers to upgrade given the new chemistry, given that these instruments are validated and some of them are using NextSeq 550Dx? Any thoughts on NextSeq 2000 Dx? Appreciate it. Thank you.
Sure. Thank you, Puneet. I got three questions. One is, give you an update on the NHS ramp and how that's looking for the back half of the year, and then also comment on how to bridge the fact that we've talked about three to 500,000 samples, whereas you have Matt Hancock talking about 5 million genomes being done. Bridge that. You also asked about NextSeq, and that upgrade cycle, especially talking about the Dx instruments and whether the Dx customers for NextSeq will be part of this upgrade cycle. In terms of the NHS, we are happy with how things are progressing, frankly. Over the last few months, we've given you the update announced at JP Morgan, that we have signed a contract with GeL to provide the lab testing services, the genomic lab testing services for the NHS. Our teams are deeply engaged.
We hosted the leadership here in California to review plans going forward. As of now, we feel very good about how things are tracking in terms of the ramp for the NHS, and we expect to see that sort of mid-year going forward. The first phase is the 300,000 scaling up to 500,000 genomes to be sequenced. We've done a bottoms-up analysis with that team on how you get there. What we've said publicly is that genomic testing is going to be a standard of care for just over 20 genetic diseases and four different types of cancers, starting in the middle of the year when we ramp up with the NHS. Standard of care for the U.K. population, 55, 60 million people. Obviously there are a lot more than 20 genetic diseases and a lot more than four cancers.
Matt Hancock is really driving the NHS to be more ambitious about how we want to roll this out for the entire U.K. population. There are not 20 genetic diseases, there are 6,000 genetic diseases. There are awfully a lot more than four cancers. As you start to do the build-up of what happens when you expand this across cancer indications and across more genetic diseases, as well as other areas that genomic testing could be helpful for better outcomes at lower cost, that's how you get to a number like 5 million. That's not yet a committed path with a committed timeframe, but that's the ambition. That's the NHS ramp. In terms of NextSeq and the upgrade path, especially around Dx.
The way we expect the DX market to play out is that if you are a NextSeq 550Dx customer today, or you are in the pipeline to buy a NextSeq 550Dx, we expect that you will either keep using the one you have, or you will buy a new NextSeq 550Dx, and that more likely than not, you won't be purchasing the 2000 or the 1000 right now. That goes to the point you made, which is, look, you have validated workflows. In a lot of cases, you may actually have a cleared assay built on our cleared box. So you'll certainly keep an eye out for when the 1000, 2000 DX instrument gets announced and comes to market. For the foreseeable future, your capacity adds will be DX boxes, NextSeq 550Dx boxes.
We don't expect those customers to be driving the pipeline for NextSeq 1000 and 2000 right now.
Your next question comes from the line of Patrick Donnelly with Citi. Your line is open.
Thanks, guys. Sam, maybe just on the guidance linearity you talked about, maybe just a bit more color there. 1Q obviously came kind of well below where the street was, only looking for 1% of growth on the revenue side, EPS, a pretty big year-over-year decline. Can you just talk through that a bit more? On top of that, on the guidance side, just the full-year visibility into some of the PopGen revenues, timing of some of those initiatives that have yet to start. Obviously, in 2019, we saw some pushouts. Can we just talk to the confidence level in the numbers, particularly that 200,000 from UK Biobank? How much can we see that shift, either upside or downside from that number? Thanks.
When we think about Q1, thanks for the question, by the way, Patrick. When we think about the Q1 number, first of all, as we look across Q1, but also across all of 2020, this is not unlike other years where we've had instrument launches. Going back to 2017, I know NovaSeq is a different instrument, but the linearity is very similar to what we saw in 2017. When we think about Q1, if you're comparing versus Q4, obviously, you have two or three big factors that drive the decline from Q4 into Q1. One is seasonality and the lower instrument placements, specifically NovaSeq in Q1 versus Q4. Seasonality across most of the business, across sequence and consumables as well.
What we also have seen in the past, where Q1 usually is a big step-up because of the DTC business and some of the processing of samples that come back after the holidays, well, that doesn't exist this year. We have, obviously, a much weaker DTC performance for both Q1 and the year. Finally, another key factor is the fact that with the launch of NextSeq 2000, you will have constrained NextSeq shipments in Q1 because customers are pausing, waiting for the new instrument to come out, and the new instruments will not be available until late in Q1. You will have much fewer NextSeq shipments overall in Q1. Those are the key factors that drive the sequential decline from Q4 into Q1.
As we think about the rest of the year, obviously, you will have the NextSeq 2000 launch that continues to ramp up. You have instrument placements for NovaSeq will ramp up across the year. You will have the population genomics opportunities that we talked about. The UK Biobank is in full production mode and will continue to process those 200,000 samples that we talked about across the full year. You have also the expectations that we shared at JP Morgan around All of Us starting in the middle of the year, the NHS Commissioning with GeL starting in the middle of the year. We have, obviously, confidence with the ramp that we expect over the course of 2020. Q1 is constrained for the reasons that I mentioned, and obviously, the NextSeq 2000 launch plays a big part in that.
It's not unlike other launch years that we've had before.
Your next question comes from the line of Dan Leonard with Wells Fargo. Your line is open.
Thank you. Just wanted to circle back on those PopGen expectations. Sam or Francis, how comprehensive is that disclosure? Meaning the three programs you flagged, the UK Biobank, All of Us, and NHS, are those the three you've gotten permission to offer some disclosure and there could be other ones that are material coming into the fold in 2020, or are those really the three that are going to drive numbers in 2020?
Maybe I'll start, and Francis can chime in as well. This is not about what we have permission to disclose. This is about we want to make sure we focus on the key PopGen projects, and those three are the key PopGen projects that we talked about last year, and we wanted to talk about this year and provide very specific visibility to them. U.K. Biobank, which has started and will continue to process across all of 2020. All of Us, which was big talking point in 2019, and we wanted to give clarity as to when we expect that to start and the number of samples associated with it.
Obviously the NHS Commissioning project, which was also a talking point last year, now we're pleased to say that we have this agreement with GeL and the NHS, and that's going to start as well. The UK Biobank has started too. There are other population genomics initiatives within our 2020 financials. We will not be specifically talking about those. Those are part of our business. Some of those have started in 2019. Some of them we expect to start in 2020. We really wanted to focus on the three key material ones.
Thank you.
Your final question comes from the line of Jack Meehan with Barclays. Your line is open.
Thank you. Good afternoon. Just maybe to round out, I was hoping, following the conclusion of the PacBio potential deal, I was just hoping you'd give us an update on Illumina's strategy and long-read sequencing. It seems like Illumina is going to need to innovate their way into that market. Maybe just what's the strategy moving forward? How big of a priority is this in terms of R&D? Can you maybe give us some comfort around the level of prioritization versus some of the other things you're working on, like clinical? Thanks.
Yeah, sure. Let me answer that, Jack. We continue to believe that the long-read market will be an adjacent market in the overall sequencing market, representing about 5% of the overall sequencing market. That for the foreseeable future, the two markets will intersect very little, to be honest. The big markets for us, whether it's oncology testing or NIPT, don't really need the long-read capability, and the big markets for long-read, de novo sequencing, new species sequencing, really need that long-read capability, short reads aren't a suited technology for that market. We continue to believe that'll be a complementary market to the short-read market, representing about 5% of the overall sequencing market. What we were excited about in the PacBio deal was that our engineering teams are fantastic at driving the price, the cost of a machine down.
We have terrific engineering and operations teams and terrific technologists. What we wanted to do was try and accelerate that market because we know that there is a knock-on effect about what's good for that market is good for our market. Once you sequence a new species, for example, you do the bulk sequencing in that species on short-read technology. We're going to continue to look for ways to move that market more quickly than it is moving now. It's a small part of the overall sequencing market. It dwarfs in comparison to the clinical market opportunity, which is much, much larger than the long-read market opportunity. In terms of trade-offs that we would do internally, it's not even close when we think about spending more to drive the oncology testing market forward, for example, versus the long-read market.
We've been public about the fact that we have our own long-read programs that we're working on internally. We're going to continue to work on those internally. Far and away, the higher priority for us continues to be the short-read core markets, both the research markets and the clinical markets.
There are no further questions at this time. I will turn the call back over to the presenters.
Thank you. As a reminder, a replay of this call will be available as a webcast in the Investors section of our website. Thank you for joining us today. This concludes our call, and we look forward to our next update following the close of the first fiscal quarter.
Ladies and gentlemen, this concludes today's conference call. Thank you for participating. You may now disconnect.