Pacific Biosciences of California, Inc. (PACB) Q3 2013 Earnings Report, Transcript and Summary

Good day, ladies and gentlemen, and welcome to the Pacific Biosciences of California, Inc. Third Quarter 2013 Earnings Conference Call. [Operator Instructions] As a reminder, today's conference is being recorded. I would now like to turn the conference over to your host for today, Ms. Trevin Rard. Ma'am, you may begin.

Thank you. Good afternoon, and welcome to the Pacific Biosciences Third Quarter 2013 Conference Call. With me today are Mike Hunkapiller, our Chairman and CEO; Susan Barnes, our Chief Financial Officer; and Ben Gong, our Vice President of Finance and Treasurer. Before we begin, I'd like to inform you that comments made on today's call may be deemed to contain forward-looking statements. Forward-looking statements contain such words as believe, may, estimate, forecast, anticipate, continue, intend, expect, plan, the negative of these terms and other similar expressions and include the assumptions that underlie such statements. Such statements may include, but are not limited to, revenue, margin, cost and earnings forecast, future revenue implied by the company's backlogs, expectations of future cash usage and other statements regarding future events and results. Actual results may differ materially from those expressed or implied as a certain -- as a result of certain risks and uncertainties. These risks and uncertainties are described in detail in the company's Securities and Exchange Commission filings, including the company's most recently filed quarterly report on Form 10-Q. The company undertakes no obligation to update, and prospective investors are cautioned not to place undue reliance on such forward-looking statements. Please note that today's press release announcing our financial results for the third quarter 2013 is available on the Investors section of the company's website at www.pacb.com, and has been included on Form 8-K, which is available on the Securities and Exchange Commission's website at www.sec.gov. In addition, please note that today's call is being recorded and will be available for audio replay on the Investors section of the company's website shortly after the call. Investors electing to use the audio replay are cautioned that forward-looking statements made on today's call may differ or change materially after the completion of the live call, and that Pacific Biosciences undertakes no obligation to update such forward-looking statements. At this time, I'd like to turn the call over to Mike.

Thanks, Trevin. Good afternoon, and thank you for joining us today. We are pleased with our third quarter results and the continued progress we're making in driving the adoption of our products. Highlights of our third quarter are as follows: we booked new orders for 5 PacBio RS II systems and installed 6 systems, which brings our install base up to 83 and our backlog of systems to 9 as of September. We booked 7 new orders for RS II upgrades, which brings our total system upgrades to 61 out of a possible 74 RS systems that were installed prior to our release of the RS II. All but 1 of those 61 upgrades were installed as of September 30. Consumable revenue for the quarter was $2.1 million, which is up 64% from Q3 of last year, and up 10% sequentially from our second quarter revenue, signaling continued strong utilization in our install base. Total revenue grew sequentially from $6 million in Q2 to $7.4 million in Q3. Last month, we signed an agreement with Roche Diagnostics to make them our partner for developing and commercializing products for the in-vitro diagnostic market, and we received $35 million in cash upfront from Roche as part of that agreement. We ended the quarter with $127 million in cash, which is up $20 million from our cash balance at the end of June. And finally, we launched our latest chemistry and software release earlier this month. This P5-C3 chemistry release features photo-protected dyes, which enables extra long read lengths. We have already seen customers achieving N50 read lengths in excess of 10,000 bases, which means that more than half the data they get from run is comprised of reads longer than 10,000 bases and the longest reads they're getting are over 30,000 bases long. In addition to these highlights, we have also seen a significant increase in the number, recently averaging 1 paper per week of customer publications based on the use of our SMRT sequencing technology. We're excited about all of these, but I will highlight a few that illustrates some of the key advantages of the SMRT technology. In particular, its extraordinary read lengths and lack of sequence context bias enable analysis of important genomic regions that are frequently missed entirely using short read technologies. Either because they cannot span DNA repeat regions or because of the use of DNA amplification biases against low complexity sequences. The first paper I'll highlight is a publication in the American Society for Microbiology Journal, Genome Announcements, entitled De Novo Assembly of the Streptomyces Strain in G1 Genome Using PacBio Single Molecule Sequencing. Streptomyces species have approximately 8.7 megabase genomes that contain numerous long repeat structures in key gene clusters and have greater than 70% average GC content. Genome assembly based on combinations of short reads, both Illumina HiSeq and Roche 454 data gave an assembly in 466 pieces with over 10% of the sequence missed entirely. Using data from the PacBio RS, the researchers were able to generate an assembly that overcame these problems and result in genomic regions key to the medical and industrial significance of these bacteria. The second publication is from Genome Biology entitled, Reducing Assembly Complexity of Microbial Genomes with Single Molecule Sequencing. The authors of this paper describe the inability of first and second generation sequencing platforms to completely reconstruct bacterial chromosomes, which has resulted in most bacterial genomes being left unfinished. In this study, the author sequenced 6 bacteria with PacBio and analyzed the repeat complexity of 2,267 other genomes. Based on the results, the authors conclude that they can assemble most known bacteria at finished grade quality without gaps using only a single library with the PacBio RS, and that accuracy of these assemblies with quality values greater than 60 is much better than what they could achieve with either short read technologies or hybrid assembly techniques. The authors went on to state that automated assembly of long single molecule sequencing data reduces the cost of microbial finishing to $1,000 for most genomes and continuing improvement should drive that cost even lower. The next publication is from BMC Genomics entitled, An Evaluation of the PacBio RS Platform for Sequencing a De Novo Assembly of a Core Plast Genome. The researchers were studying the genome of a close relative of strawberry plants. This chloroplast genome is relatively low in GC Content, with an average of 37%, and has several very long and almost identical repeat structures. Using the PacBio platform, the researchers were able to obtain the full sequence of the 155 kilobase genome in a single contig from as little as 35x coverage. In contrast, using high speed data, up to 9,100x coverage, they were only able to get an assembly with 7 contigs with almost 10% of the sequence missing entirely due to very low GC content in the repeats. The last publication I'll highlight is one that appeared in the American Journal of Respiratory Cell and Molecular Biology entitled, Genome Reference in Sequence Variation in the Large Repetitive Central Exon of human MUC5AC. Last quarter, I mentioned that researchers at the University of North Carolina were working on studying samples from cystic fibrosis patients to better understand the relationship between expanding repeat regions in this gene and the severity of disease. This publication highlights some of that work and describes the region as one that contains the gap known in the human genome reference comprised of a highly repetitive and complex central exon. This exon region has not been successfully sequenced in the past by any other sequencing technology. They noted that the entire region was in general disarray prior to using SMRT sequencing. In the paper, the researchers described how they used SMRT sequencing to obtain long sequence reads and robust coverage that allowed for de novo sequence assemblies spanning the entire gene. They also identified genetic variation in the 4 patient samples that were studied. The study illustrates and the authors concluded: "That SMRT sequencing can be used successfully to perform de novo assemblies of large repetitive sequences to fill in gaps in important regions of the human genome." This past month, we attended the second Annual International Conference on Genomics in the Americas, which was held in Sacramento. There were several presentations at the conference highlighting PacBio technology. In the keynote presentation, Nobel Laureate, Rich Roberts, described his recent work on bacterial methylomes using the PacBio RS II. His team at New England Biolabs has been studying methyltransferase systems to better understand how certain bacterial organisms evolve and acquire pathogenic genes. For instance, his team used the PacBio RS to study the epigenetics of the E. coli strain responsible for the outbreak in Germany in 2011, and found that it was missing genes found in other bacterial strains, which restrict incoming DNA that is methylated. Without that protection, the strain had picked up the phage with the gene that encoded shiga toxin, causing it to be pathogenic. Another presentation I'll highlight was made by Ken Dewar from McGill University in Montréal. His talk focused on the analysis of complete bacterial genomes in less than 24 hours in response to outbreaks of infectious diseases. His team at McGill University has developed protocols to rapidly sequence and analyze outbreak samples using the PacBio RS II platform. Dr. Dewar highlighted the PacBio RS II's ability to generate high-quality, complete bacterial genomes in a very short timeframe. McGill also uses the Illumina MiSeq for studying bacterial genomes. However, while the MiSeq data is useful, McGill cannot produce completely finished genomes with the MiSeq data alone, due to repetitive elements in the bacterial genomes, which were not spanned by short read sequences. Dr. Dewar concluded that with sequencing cost coming down from product improvements and streamlining protocols, sequencing is becoming an attractive option for outbreak monitoring. The last presentation I will highlight was from our Chief Scientific Officer, Jonas Korlach, entitled Finished Microbial Genomes and Epigenomes on a Large Scale. Jonas described the results from a joint study with the Centers for Disease Control, Cornell University, New England Biolabs and the University of California at Davis, involving the study of the food-borne pathogen Lysteria monocytogenes. In this study, the researchers were able to easily assemble 16 different strains down to single contigs or approximately 3 million bases each. Furthermore, they discovered mobile elements in certain strains, which should only be identified using SMRT sequencing. Finally, when looking into the strain's epigenomes, they identified modified bases that are likely to have regulatory roles in the expression of the genes that cause the strains to be pathogenic. This past month, we hosted a West Coast user group meeting, UGM, and a bioinformatics training session here in the Bay Area. We had 85 attendees at the UGM and 36 attendees at the training session. We found these types of meetings to be very effective in sharing the latest customer accomplishments and sequencing approaches in an informal fashion that is much closer to realtime than peer-reviewed publications. At the UGM, there were 12 diverse customer presentations covering a wide range of sequencing applications with the PacBio RS II platform. I'll highlight 3 of these presentations. The Joint Genome Institute has been and continues to be a leader in microbial sequencing with their PacBio RS II systems. For the past few months, they have assembled genomes using between 3 and 10 SMRT cells each. And now, they are down to using just 2 SMRT cells each. Alex Copeland at JGI presented an interesting historical recount of their cost to sequence bacterial genomes over the past decade. 10 years ago, when all they had was Sanger sequencing, they would spend up to $85,000 to sequence a genome to a high level of quality. In order to reduce cost, they adopted the use of second-generation systems, such as HiSeq, to sequence many more genomes at a lower cost per genome but the drawback was that the genomes were lower quality and never finished. Now with PacBio, they were able to produce high quality finished genomes at low cost. Furthermore, with all the improvements and throughput we have delivered over the past year, the team at JGI has expanded their work with the PacBio platform in the fungal and plant genomes, which are one to several orders of magnitude larger than microbial genomes. At the Salk Institute, scientists have been studying Arabidopsis thaliana, which is one of the model organisms used for studying plant biology. Its genome consists of approximately 120 megabases and researchers have spent millions of dollars attempting to assemble it. Dr. Luo from the Salk Institute shared some of the early analysis of the genome using SMRT sequencing and the results were very impressive. As example, previous assemblies with short read technologies combined with Sanger sequencing yielded contigs or pieces with a maximum length of 460,000 bases. In comparison, the assembly with PacBio alone yielded contigs with a maximum length of close to 13 million bases, or a 28x improvement. The contig N50, which describes the contig length at the halfway point of the distribution was only 66,000 bases for the reference short read assembly compared with 6 million bases for the PacBio assembly, which represents a 93x improvement. It's results such as this that we believe will lead to much more work in the area of plant genomics with their technology. Mount Sinai presented some of their latest data on a variety of projects ranging from studying the spread of MRSA infections in the hospital to sequencing human genes with disease association. With respect to monitoring MRSA, they have created a pipeline using PacBio to generate data from a patient sample in approximately 48 hours at a cost of less than $300. In the area of human sequencing, they are focused on previously ignored regions of the genome that have not been well fed in the past because other sequencing technologies cannot read through them. In particular, they are studying genes with complex tandem repeat regions, which are associated with diseases such as hemiplegic migraines and episodic ataxia. And now, I'll highlight our most recent product announcement, the P5-C3 chemistry and software enhancement, which we released earlier this month. This is our latest in a series of product enhancements over the past 2 years that improves the performance of the PacBio RS II platform for both existing and new customers. We are committed to systematically delivering product improvements in short timeframes, and this allows our customers to take full advantage of their PacBio equipment. We believe this latest release will have a significant impact in our customer's ability to assemble larger-sized genomes and to close existing gaps on partially completed genomes. For example, Mount Sinai is one of our customers that had early access to the P5-C3 chemistry. At our UGM, they shared some of their latest results using a combination of 20,000 base libraries and our P5-C3 chemistry. They reported achieving read length N50 of 12,500 bases and their longest reads were over 34,000 bases. Our latest release also includes 20,000 base size selected library protocols optimized for use with the Sage Science and BluePippin DNA size selection system, along with an upgraded version of SMRT analysis, which introduces an update to the popular HGAP assembler. The new HGAP II assembler significantly decreases the time required to perform the noble genome assembly and extends the range of genome sizes that can be assembled. Additional software features include deployed aware set detection, as well as haplotype phasing of amplicons, which provide users the opportunity to utilize the platforms for applications such as HLA analysis. Now I'll make a few comments on our partnership with Roche, which we announced last month. As I mentioned on a recent conference call, we believe this partnership accelerates our ability to participate in a large and growing molecular diagnostics market. Roche is the world leader in in-vitro diagnostics. They bring a wealth of knowledge and experience in bringing products to market in this space. They have the sales and marketing infrastructure to accelerate sales once the products are available. We received an initial $35 million in funding from the agreement, and we expect to receive an additional $40 million in the future, provided that we achieve certain development milestones. Our development activities are up and running and while we do not plan on reporting on the details of the development activities, we look forward to updating you when we reach significant milestones. Now switching back to our commercial activities. During the third quarter, we were pleased to see our consumable revenues growing despite the impact of summer seasonality. Consumable revenues were up 64% year-over-year and up 10% from last quarter. We believe this is a strong testament to the value we are bringing to our customers. As we mentioned in the past, we believe that strong customer references fuel ongoing consumer purchases. On the capital sales side, our instrument bookings for the quarter were 5 units, compared with 4 units in Q3 last year, and compared with 7 in Q2 of this year. While we were a bit disappointed with the sequential decrease, we continue to have a healthy pipeline. And for the year, we continue to target a doubling of total 2013 bookings compared with our bookings from last year. That said, I would like to briefly comment on how the U.S. government funding environment is impacting our short term look at our business. A number of our existing customers in the U.S. are either federal agencies or laboratories funded by government grants or contracts. When the government shut down this quarter, a number of our U.S. customers were forced to temporarily shut down their systems while their labs were closed. Many government scientists were not allowed to come to work or even check their e-mail. This has impacted our current quarter consumable sales and has disrupted the dialogue we're having with certain government agencies concerning future Instrument business. Therefore, we are being somewhat cautious about providing near-term forecast until we better understand the full implications of the recent shutdown and ongoing U.S. budget negotiations. We still feel positive about the overall trajectory of our business. Finally, many of you are aware that the annual American Society of Human Genetics Conference is being held this week in Boston. For those of you attending, we are hosting a lunchtime workshop at the conference this Thursday, entitled Characterizing Structural Variation in the Human Genome using Long Read SMRT Sequencing. The workshop will include 3 presentations that illustrate the use of SMRT sequencing to characterize important and complex genome regions that have been refractory to satisfactory analysis by short read technologies. We hope to see some of you there. With that, I'll turn the call over to Susan.

Thank you, Mike, and good afternoon, everyone. I will begin my remarks today with a financial overview of our third quarter that ended September 30, 2013. I will then provide details on our operating results for the quarter with a sequential comparison to the second quarter of 2013, as well as the year-over-year comparison to the third quarter of 2012. I will conclude my remarks with a brief discussion of our balance sheet. Starting with our third quarter financial highlights. During the third quarter, we recognized revenue of $7.4 million, and incurred a net loss of $20.5 million. We ended the quarter with $127 million in cash and investments, an increase of $20 million from the balance of the end of Q2. The Q3 ending cash balance includes proceeds of $35 million from our recently announced agreement with Roche, offset by cash used during the quarter of $15.1 million. Breaking down our revenue. Total revenue from the quarter was $7.4 million, up $1.4 million from the $6 million of revenue recognized in Q2, and up $4.6 million from the 20 -- the $2.8 million realized in Q3 of 2012. For instrument revenue, in Q3 of 2013, we recognized $3.7 million compared to $2.7 million in Q2. A year ago, in Q3 of 2012, we had no instrument revenue. Consumable revenue was again strong this quarter. Consumable revenue for the quarter totaled $2.1 million, up 10% from the $1.9 million recognized in Q2, and up 64% or $800,000 from the $1.3 million recognized in Q3 of 2012. Service revenue increased, growing approximately $200,000 sequentially from the $1.4 million in Q2, to $1.6 million in Q3 2013, and up approximately $300,000 from the $1.3 million in Q3 of 2012. And finally, as was the case last quarter, we did not have grant-related revenue in Q3 2013, compared with $200,000 in Q3 of 2012. Gross profit for the quarter was $1.2 million, representing a gross margin of 17%. This is approximately $200,000 higher than the gross profit recognized in Q2, and $1 million higher than the $200,000 we recorded in Q3 of 2012. Moving to operating expenses. Operating expenses in the third quarter totaled $21.2 million, which included $2 million of professional and advisory expenses related to the Roche agreement. Spending in Q3 was relatively flat to the $21.1 million incurred in the previous quarter or $1.6 million lower than the $22.8 million incurred in Q3 of 2012. Breaking down our operating expenses, R&D expenses in the quarter were $10.4 million, down $1.3 million from the $11.7 million recognized in Q2 of 2013, and $2.2 million below the $12.6 million recognized in Q3 of 2012. Lower expense in Q3 2013 is related to timing of certain research and development activities, as well as the realization of continued operating efficiencies in the group. R&D expenses for the quarter includes $800,000 of noncash stock-based compensation expense. Sales, general and administration expenses for the quarter were $10.8 million, a $1.4 million higher than the $9.4 million incurred in Q2. This increase was the result of $2 million of expense incurred in association with the Roche agreement. Year-over-year, sales, general and administration expenses increased by $600,000, from $10.2 million in Q3 of 2012. This increase, again, was driven by the expense associated with the Roche agreement but offset by continued operating efficiencies realized in the group. Ben will provide further guidance on our ongoing expense rate later in the call. SG&A expense for the third quarter includes $1.2 million of noncash stock-based compensation expense. Also in the area of other income and expense, this quarter we recorded approximately $600,000 of interest expense associated with our debt, which is similar to what we recorded last quarter. Now turning to our balance sheet. Cash and investments totaled $127 million at the end of the third quarter, up $20 million from the previous quarter. This increase in our cash balance includes proceeds of $35 million from the agreement we entered into with Roche. Excluding the proceeds associated with the Roche agreement, cash used in the quarter was $15.1 million, reflecting our second quarter net loss of $20.5 million, less $3.5 million in noncash expense from stock-based compensation expense and depreciation, as well as cash contributions from various balance sheet changes. Among other accounts, accounts receivable decreased $300,000 from $4.1 million at the end of Q2 to $3.8 million at the end of Q3. And inventory balances decreased $500,000 from $10.3 million at the end of Q2 to $9.8 million as of September 30, 2013. In closing, on the balance sheet, I would like to point out that our agreement with Roche does not contain any equity component. This concludes my remarks on the financial results for the quarter and I would like to now turn the call over to Ben.

Thank you, Susan. I'll be providing updated forecast on our 2013 financial performance. Starting with bookings, as Mike mentioned earlier, we're continuing to target a doubling of our 2012 bookings. However, with ongoing uncertainty surrounding government funding, we may incur delays that could impact our fourth quarter bookings. Moving on to revenue. Our third quarter revenue results were in line with our expectations. We installed 6 systems in Q3, which is consistent with what we had anticipated. We had started Q3 with 10 systems in backlog, adding the 5 new bookings and subtracting the 6 installations from the third quarter, we ended the quarter with 9 systems in backlog. Looking forward to Q4, we anticipate installing a majority of these systems during the fourth quarter, as we coordinate with customers who are preparing their sites for installation. Now regarding consumables, our Q3 revenues were higher than we had forecasted. We are pleased to see sequential consumable revenue growth last quarter, despite the seasonal summer slowdown. Looking forward to Q4, unfortunately, the government shutdown has had an impact on our consumable sales, as several of our busier customers were not permitted to run their systems during that time, and they have spent some additional time bringing their systems back up. Normally, we would forecast sequential growth in consumable revenues for the fourth quarter, but due to the shutdown, consumable revenues may turn out to be relatively flat compared with Q3. However, assuming no further significant interruptions to our install base, we expect to see sequential growth in consumable sales in subsequent quarters. Finally, as part of our agreement with Roche, we received a nonrefundable payment of $35 million in September. We recorded this payment as deferred revenue on our balance sheet, and we expect to amortize this into revenue on our income statement in future periods. For the fourth quarter, we anticipate recognizing approximately $1.5 million of the $35 million Roche payment into revenue. Excluding the revenue benefit in the fourth quarter from the Roche agreement, we continue to expect to record higher total revenue for 2013, compared with our revenue for 2012. With regard to gross margins, we have maintained our gross margins at the mid-teens as expected. Going forward, our gross margins will likely benefit from the revenue we recognized from the Roche proceeds, as those revenues are expected to be recorded at 100% margin. Our operating expense for Q3 came in at $21 million, including approximately $2 million in expense related to the Roche transaction. Excluding those transaction-related expenses, we reduced our operating expense to approximately $19 million in the quarter. As Susan mentioned earlier, a significant portion of the reduction was due to timing of certain research and development activities. We continue to monitor expenses closely, and subject to timing differences, we expect to maintain our quarterly operating expenses between $20 million and $21 million. Please note that our operating expenses have included noncash, stock-based compensation expense and appreciation expense that together amounts to approximately $3.5 million to $4 million per quarter. As Susan mentioned earlier, interest expense for the quarter included about $600,000 related to our debt financing, which was comprised of $450,000 in cash interest, plus approximately $170,000 of noncash amortization. We expect to record similar interest expense for the fourth quarter. With regard to cash usage, we have previously forecasted using $70 million or less in cash this year. For the first 3 quarters of the year, we have only used $49 million. By lowering our cash burn rate, we now expect to use approximately $65 million in total for the year. In comparison, we had used approximately $77 million in cash during 2012. We have focused on reducing our cash burn by driving higher top line revenues and controlling expenses and we intend to maintain this focus going forward. With the recent proceeds we received from Roche, we expect to end the year with at least $110 million in cash. And with that, we will open the call to your questions.

[Operator Instructions] Our first question comes from Bryan Brokmeier from Maxim Group.

Since you signed the Roche agreement, have you recognized any increase in inquiries or interest from researchers outside of diagnostic field?

Well, the answer is yes, particularly people who had used the Roche 454 technology, which up until PacBio came along, was the longest read technology in any of the second-generation systems. I think the other thing it has had is some increased confidence in our customer base and our potential customer base just because of the sort of endorsement from a major diagnostic player like Roche as well, which spreads not just to the diagnostic world, but to the research community, which is mostly where Roche has sold their sequencing technologies up to now.

And now that you significantly shored up your cash position, do you anticipate that you'll continue to utilize your ATM, raise more cash?

Well, Brian, as we had mentioned in our call about a month ago, with the Roche proceeds and anticipated proceeds from some of those milestone payments, we think we have enough cash that will last us until 2016. So we're doing pretty well in terms of runway. We don't give any forecast on uses of the ATM. So we'll just kind of leave it at that.

And beyond the fourth quarter, is that -- was it -- would you say, $1.5 million in the fourth quarter, of that $35 million, is that the run rate that we should expect in future quarters, about $1.5 million throughout 2014 as well?

Yes, that would be a fair estimate to make, just sort of $1.5 million on a quarterly basis.

And then, lastly, can you provide any update on 100,000 genome project. How many genomes have been run so far in PacBio system or any other details you can provide?

Well, this is on the bacterial genome project. No, they'll put out an update as they do them. I mean, recognize that a lot of those labs were government labs that were shut down in the first 3 weeks or so of the quarter, effectively. But they will put out their own data release as they both do the work and submit the data into national databanks.

The only thing I'll add on that, Brian, is we had mentioned that the FDA ordered a second system in the second quarter, we did install that in the third quarter, they got up and running, but then, as you know, then they had to shut it down during the government shutdown. So that's one of the sites that we had mentioned as one of our busier sites that was disrupted.

Our next question comes from Bill Quirk from Piper Jaffray.

First question for me. I'm just trying to make sure I get my numbers to tie here correctly. I think, I heard, Mike, you said in the prepared comments that you upgraded 7 systems in the quarter, and if so, if I'm doing the math correctly here, it looks like the average instrument price was off a little bit. Can you just elaborate on that, was there something -- somewhat akin to what we saw back in the fourth quarter of '12, where it was just a geographic mix?

Yes, this is Ben, I'll try to give you a little bit more hint on that. So we took 7 new orders on the upgrades, but we actually installed quite a few. So Mike also mentioned that of the 61 that we received so far, we've now installed a total of 60. I believe we installed at least, I think, 25 or 26 of those in the third quarter. Now -- but your question, I think, has to do with ASPs, and as we said in the past, the average selling price each quarter moves around based off of the geographic mix, and so without getting any specifics on our pricing in each of these places, generally speaking, a higher mix of European euro base sales might make the ASPs slightly higher, whereas in quarters where you have not installed so many in Europe, then it might be lower, and that's causing most of the fluctuation that you see quarter-to-quarter.

Okay. Understood. To be fair, Ben, if I just -- again, if I'm quickly doing the math here, it looks like a pretty significant swing quarter-to-quarter though, that's just all mix?

Well, you also had fewer upgrades in Q3 versus Q2. So you have a couple of things going on there. So you have fewer contribution from upgrades and then you had a different mix on the systems that we installed. And just keep in mind, in your denominator in Q3 was only a 3, so when you're doing numbers that are that small, sometimes you get, again, average numbers that can really fluctuate.

Fair enough. And then, Mike, just to elaborate on the comments referring to some of the challenges that some of your customers are having obviously, you've specifically called out FDA in having to shut down their system. Can you just kind of parlay that into your confidence regarding hitting the bookings number guidance for the year. I mean, I realize they're not completely intertwined, but to the extent that people had to shut down operations, are they immediately back and looking to buy equipment?

Well, the biggest issue we had with the shutdown per se was with the ongoing consumables because you basically just lost completely 2.5 weeks, and then, as Ben pointed out, it takes them a little bit to kind of get their lab back and running again, and we haven't completely factored that into exactly what we expect. It didn't obviously affect our whole customer base but it did impact a disproportionate share for us because we have a pretty good penetrance into a lot of the government labs, particularly in the microbial and viral space. The issue we were trying to highlight was that, clearly, there are a lot of budget negotiations going on. And that does impact a lot of these agencies' ability to go ahead and pull the trigger on additional purchases, as they're trying to figure out what their budget's going to be as opposed to the sort of month-by-month or I guess a couple of months by month ongoing -- continuing resolution. So our anticipation is it could slow it down, it might not, in terms of them being able to make decisions, and so we're just being somewhat cautious because we have little control over the machinations going on in Washington at this point.

Our next question comes from Tycho Peterson from JPMorgan. Tycho W. Peterson - JP Morgan Chase & Co, Research Division: Going back to the first question on the Q&A on just kind of Roche, subsequent to your announcing the deal, they did announce plans to exit 454, so maybe can you talk a little bit about plans to go after that install base, obviously, you've been under some pressure for a while, and does having a relationship with Roche help you out at all in that process?

Well, it can't hurt. No, I mean, they just officially made that announcement public really in the last week. But that said, I think, there was an article in genome Web or in sequence, I forget which, which highlighted the fact that several of those customers are now already actively thinking about what they're going to do next in terms of the technologies that they would want to employ, and given the fact that a lot of them bought the Roche technology because of the longer reads that it had versus some of the others, that certainly gives us a position to go on and talk to these people. That said, it has some -- been something that's been on our radar in the past and we've done some modest focus on that, particularly in Europe where they were strongest. But it's a week-old news for most people. So we are in the process of figuring out how rapidly people want to begin to switch over. And we've got a pretty good positioning relative to the 454 technology because, obviously, we have much longer read lengths, our throughput is now pretty much equivalent per cell to what they get in the run with their technologies at much, much lower cost. So for people who need the longer read lengths, even at the length defined by the Roche 454 system, we think we have a pretty strong argument. Tycho W. Peterson - JP Morgan Chase & Co, Research Division: And you guys showed a nice sequential uptick in consumables, up by 10% as you called out. Can you maybe just talk to some of the underlying trends that are driving that and how concentrated was the uptick among your customer base?

I'll take a stab at it, and then, maybe Ben can add some color as well. I think it was pretty much across the board. We certainly got an uptick at a broader set of customers than maybe we had at the beginning of the year, which has been good. But I don't know that there was any particular customer that waited the whole increase versus that. I think, we were expecting it to be relatively flat seasonally, and in the summer, you get some slowdown, typically in consumables use in the U.S., down a lot. You tend to have a good part of the European base on vacation for 1/3 to 1/2 of the quarter, and we were pleasantly surprised that despite all that, which I'm sure it was going on, we still saw that sequential increase. So I've seen that as a positive sign.

Yes. I would say that there's not any 1 or 2 things that stand out there. It was based off of the install base because, again, there was only 3 more systems in the install base, which is not that much comparing 1 quarter to the next. And September was a pretty good month. So it's always a little bit of a guesswork how good you're going to do at the end of the summer, and September turned out to be pretty good. Tycho W. Peterson - JP Morgan Chase & Co, Research Division: And I guess, what I'm kind of getting at is as we think about the utilization picking up, are you starting to have more discussions with customers about adding additional systems or enough of them not yet pumping into that kind of $300,000 threshold on consumables yet?

Well, we mentioned one that, that was the second purchase this quarter and we had one on order before then, so.

I don't know if you mentioned that. But actually, one of the bookings in the third quarter was the second system for another site. Tycho W. Peterson - JP Morgan Chase & Co, Research Division: Okay. And then, last one, just on the technology development path, you talked about over 30,000 base pair reads with some of the new technologies, just how are you thinking about the market for super long reads, if you will?

Well, there's a super long reads but then there's just the average read going up in length, too. So once you get the really long reads, you pick up the ability to do more and more complicated genomes directly. And I think we hadn't mentioned it yet, but we posted on our blog site today, release of 10x coverage of a human-derived cell line that, for technical reasons, is actually a haploid human genome to allow people in the informatics world to begin to focus on how to deal with that large and complex genome using PacBio-only data. One of the things that you might see highlighted on Thursday afternoon by one of the collaborators there. And so, it just allows you, as you continue to push things up, to deal with more and more complicated genomes and to deal with them even better. And particularly, when you retain the lack of sequence bias, which you really cover the whole genome in these things more easily, that the combination of those 2 factors starts to make a big difference. The human genome is complicated as it is for all kinds of medical reasons, it kind of dwarfs in comparison to some of the plant genomes that are out there. And so the data that we have released last quarter on the Arabidopsis is a model organism that kind of illustrates how well you can do chromosome assemblies with long read data. And that was done, I believe, with the P4 chemistry. So as you get longer and longer reads, it's just easier and easier to deal with these things. We had assemblies there of entire chromosome arms in a single contig using only the long read data, which is just an anomaly, really, to look at genomes like that.

Our next question comes from Daniel Brennan from Morgan Stanley.

Just was wondering on the user group meeting and the training session, which seems like it was very well-attended, do you have any insight on kind of a number of those users in training session were for current owners? Were they all current owners or does that also make for potential pool of candidates to purchase new systems?

That's a good question, I'm not sure I know completely the answer. But surely, a lot of them are current customers but a lot of them were customers of customers. Because we had several instruments in what I call core facilities and their customers are people who provide them with samples and the core lab frequently gives them the raw data from the instrument and it's up to the submitter of the samples to do the secondary analysis to figure out what that's telling them biologically. And we've seen an increase in the number of people there who want to use the secondary analysis software, even though they're not yet owners of the PacBio RS themselves. And so we make that software available to them, they want to download it and use it, and they need training just like the primary users of the instrument do.

Okay. Great. And Mike, more recently, today, given your sales funnel, any type of color you can provide beyond the fourth quarter regarding the funnel?

Well, we continue to see, I think, an increase in the pipeline, the funnel. It's layered on top of the bigger budget issue in the U.S. And most of our customers are either research or they're government labs doing research for specific kinds of projects that are close to that. And so, we still, particularly in the U.S., are dependent on federal government grants for a majority of our customers, and we've seen more interest on the commercial side, but that's a relatively newer group of leads for us. But it is an area in the U.S. where given the sort of ongoing struggles, we started to spend more focus on. But that said, the pipeline of people who applied for grants and are waiting for decisions are government agencies that are trying to get additional purchases, in some cases, they're new ones and others into their budgets is actually pretty strong.

Sorry, if I can sneak one more, and I apologize. Maybe can you just remind us on the key new technology improvements you expect or customers should expect over the coming, say, 6 to 12 months, Mike?

So we would expect to continue to increase the read length, and maybe even the raw read accuracy on the really long reads on average. We would expect to continue to develop the software, particularly from a perspective of speeding up assemblies, the de novo assemblies of bigger and bigger genomes. We would expect to sort of broaden out the application base, areas of looking -- using a technology to do what's called cDNA transcriptome analysis. We didn't mention it because I had lots of other papers to mention but we've had several publications very recently of people applying the long read technology to doing analysis of RNA expression from a perspective of looking at alternative splicing events, which mix of exons are actually used to produce proteins in ways that they couldn't get from the short read technology, unlike before. Certainly, we expect it to get in the areas where we're using the ability to do long sequences to look at amplicons that look at a whole gene, a collection of exons and introns, to be able to phase variants across the various exons in a way that's not so easily done with the short read technologies. And the sequencing does that today. The software has to be very specific to be able to analyze all that. And we would expect to be putting out additional software upgrades that allow people to take more and more advantage in a more diverse set of applications. It's generally increasing the throughput by read length and number of GMWs that are loaded productively, along with a lot of work on sample prep upfront and analysis software downstream.

[Operator Instructions] Our next question comes from Amanda Murphy from William Blair. Amanda Murphy - William Blair & Company L.L.C., Research Division: So I just had a question about sales cycle, I know it's a bit difficult, given the budget issues, but what are you guys seeing in terms of timing to convert potential sales in your pipeline into sales at this point?

Anywhere from a month to a year. I don't know that we have enough numbers to give you a trend on that. I think it's better in the sense that we do see some that are fairly short. But ones that are dependent on grants in particular are government budgets, particularly in the U.S., are probably the most unpredictable at this point. NIH was scheduled, for example, to do a round of study sections to begin in October, as they normally do, and those were postponed and will be reset. We don't know what impact that's going to have on the grant, not just approval cycle, it wouldn't have been for this quarter per se anyway. But disbursement of funds in early next year. But outside the U.S., things actually are a little different than that, they're better in some cases, so it's kind of a weird mix. Amanda Murphy - William Blair & Company L.L.C., Research Division: As so that was going to be my next question, just in terms of sort of o U.S. versus U.S. split between both the instruments that you shipped this quarter and your backlog and pipeline, is there any perspective you can give on -- obviously, U.S. sales are probably more difficult these days, but just curious how much of it is international at this point?

Amanda, the install base is roughly 1/2 in the U.S. and maybe 1/4 each in Europe and Asia. And Mike's already said that in terms of the bookings activity, in the U.S., it's just been a little bit challenging because of the government funding stuff.

From a usage perspective, I think, it's been up everywhere. Amanda Murphy - William Blair & Company L.L.C., Research Division: So is it fair to say that the majority of your bookings this quarter then are o U.S.?

We don't break it out that specifically. But anyway, when we're giving cautionary comments about bookings, it's mostly centered around uncertainty on the U.S. Amanda Murphy - William Blair & Company L.L.C., Research Division: Got it. And then, just more broadly, just wondering about R&D and plans for any hardware or instrumentation changes at this point, I know it's -- you've got the agreement with Roche and there's probably some work going on there, but just curious about broader instrumentation-type spots?

Well, we're glad you're curious. No, I think the focus is on understanding in detail what it's going to take to develop a system that's appropriate for the in-vitro diagnostic market with Roche. And that's where the bulk of our efforts in that regard will be for some reasonable period of time. Appreciate that we're not going to comment in detail beyond what we've done on the timing of that and how that might impact our other overall system development plans, but that's kind of where we are. Amanda Murphy - William Blair & Company L.L.C., Research Division: Great. Okay. Got it. But in terms of the platform sort of excluding the Roche side of things, is there anything you need to do from a hardware standpoint to continue to roll out improvements on the consumables side?

Not sure how those are connected but no. So what we are trying to do on the RS platform is to continue to increase the number of pieces of DNA you can interrogate at once. We have a program with -- we've announced in the past with a European company, Imec, to work on a much higher density chip. And as we develop that technology, that's clearly something that we wanted to provide to the customers. So it's a combination of working on the multiplexing that you can get, which has a direct impact on throughput. We did 1 upgrade to the system this year in that regard, as well as increasing the performance of the chemistry, and we expect -- and the software. And we expect those factors working together will continue for some period of time to give us a lot of legs on the RS platform in terms of throughput and value improvement.

Our next question comes from Zarak Khurshid from Wedbush.

So with a large competitor out there talking about hitting its synthetic long read product for customers, how do you view that as a potential threat, or not?

Well, I don't want to take the technology lightly, given the fact that it comes from a very successful company. That said, in looking at the molecular technology, we've compared the results that they've published on their website for an organism very similar in size and composition to arabidopsis. That organism, in their case, being drosophila, which is almost the same average GC content, even somewhat smaller genome, but not a lot. And what we see is exactly what we expect, that for certain regions that are relatively low in complexity, in terms of not having a lot of repeats and not having a lot of high GC or low GC content, the technology works great. Across the board, it doesn't work so good. And the problem is, it's an amplification-based approach to getting synthetic long reads. And the more amplification you do, the more bias you get, so what you see in their data is that once you get below 30% GC content, your yield or ability to see a sequence at all starts to disappear. And so you wind up increasing, in one sense, a lot of the gaps that you get. And if you look at the published data, the increase in contig length that they get by adding molecular data to their traditional mate-paired ends and so forth, regular luminous system, isn't that substantial. And so it really doesn't solve the problem at all. In some cases, it makes it harder to get certain regions out there. And if they come up with any magic bullets to make long PCR work or to get rid of the bias, that will help them. But as yet, I think, almost to a person, people that we've talked to or looked at the technology, recognized that it doesn't do a lot for them in that regard. And it certainly compares to the kinds of de novo assemblies that you can get on those, even much more difficult regions using our technology. It doesn't deal, for instance, with the whole repeat issues. Even when they do the synthetic reads, they frequently can't assemble those 8kb synthetic reads into a single contig because of the repeat problem. And so it fundamentally doesn't solve the problem because it's still stuck with short read sequences in order to generate the data. So they will be out there and it's a technology that very similar approaches have been used and some of the big laboratories have tried for a long time, several years, and it just hasn't made it into mainstream. And our reads continue to get longer and longer, they're already longer than the best you can do with that technology. And it's most of our read as opposed to a small number with a very expensive extra step in the sample prep.

Got it. And then, one of the questions earlier on the 454 business, just wanted to follow up on that, so do you have any sense for how large that business is in terms of installed bases, installed bases instruments and the average or the annual revenue contribution?

Well, you mean, from the consumables in the latter part? I don't because I don't think they published that. I think, in terms of number of units, I don't know that they've publicly said that number at all but I've seen estimates that are probably 500 to 600 installed units. Well, they've been out there for almost 10 years now. They're, as I said, they're weighted towards Europe, which is where they were strongest early, relatively weaker in the U.S., and they've done a fairly good job in Asia. But their strength has been in Europe, I think.

Operator, do we have any more questions?

I'm showing no further questions at this time, and I would like to turn the conference back to Mr. Mike Hunkapiller for closing remarks.

Thanks. In closing, we remain steadfast in our commitment to bring the unique advantages of our SMRT technology and products to our customers and the scientific community in general. We continue to make progress and we see momentum building in our business. With further product and software releases on the horizon and our new partnership with Roche in place, it will be exciting to see how this impact -- this will impact our business going forward. Thank you for listening in, and we will talk again in 3 months time.

Ladies and gentlemen, thank you for your participation in today's conference. This does conclude the program, and you may all disconnect at this time.