Helicos BioSciences, facing dwindling cash reserves, has been testing its Helicos Analysis System for new research applications, is further developing the technology, and is working on ways to place more units at more early-adopter labs.
During its first-quarter earnings call last week and a presentation at the Cambridge Healthtech Institute Next-Generation Sequencing meeting in San Diego the week before, the company gave an update on its research, technology development, and commercial activities, and talked about plans for improving its cash position.
On the commercial front, Helicos is working on winning additional early adopter customers for its Helicos Genetic Analysis system, which includes the HeliScope sequencer.
Its first customer, Expression Analysis, which provides genomic services to the pharmaceutical industry and other customers, received the Helicos instrument in March (see In Sequence 3/11/2008). The instrument is now up and running and Expression Analysis is currently testing it with standard templates, Helicos Chief Operating Officer Steve Lombardi reported during Helicos’ first-quarter earnings call last week. In March, Expression Analysis said that it anticipates to begin offering limited services on the instrument in the third quarter.
Helicos does not expect to book revenue for the Expression Analysis sale until the second half of the year, Lombardi reiterated, noting that the payment is tied to customer acceptance based on in-house instrument performance. The instrument’s list price is $1.35 million, Helicos has previously said.
Lombardi reiterated that Helicos is focusing on 30 to 40 institutions as its first customers, more than half of which he said are located outside the US. Since the Advances in Genome Biology and Technology conference in February, where Helicos scientists presented data on BAC sequencing and small RNA tag sequencing, the company has “met with decision makers” from nearly all of these target customers, according to Lombardi. He noted that “the majority” of them have visited Helicos in recent weeks to explore its technology further.
In response to an analyst’s question about whether the company has “flexibility” to discount the instrument for early adopters, Lombardi said that “you have always flexibility when you are selling capital equipment, and you are always looking at a whole host of issues to be able, in a sense, to create a win-win situation for the company and the customer.”
Asked by another analyst why the large genome centers have not yet brought in the HeliScope sequencer, Lombardi indicated that the instrument does not yet meet their performance requirements. Helicos decided long ago to focus the initial version of its instrument on candidate genome resequencing and gene expression, he said, adding that this “is not the primary charter of genome centers.”
However, since one of their charters is to investigate new technologies, “we continue to have very active discussions with them … and feel very confident that they remain a key part of our [customer] pipeline,” he said.
He added that Helicos believes “the trajectory path of our technology … is absolutely in line with the needs and the interest of the genome centers going forward. We are confident that we are going to be able to meet their needs.”
In the meantime, the company is focusing on other early adopters. “Our sales team is actively turning the interest from these important customer visits … into business,” Lombardi said, adding that the company is “deep into the sales cycle with a number of key customers.”
One institution considering buying a Helicos platform is the RIKEN Yokahama Institute in Japan, In Sequence has learned (see Transcript in this issue).
Lombardi also talked about how Helicos could improve its cash position.
The company will have enough cash to last a little more than nine months based on current cash burn rates, but revenues from instrument sales or other sources would allow it stay solvent into next year.
During the first quarter, Helicos burned through approximately $14.7 million in cash, leaving it with $37.9 million as of March 31 (see In Sequence 4/29/2008).
The company, which went public a year ago, currently has a “core cash burn” of approximately $3.6 million per month, or $10.8 million per quarter, from recurring SG&A and R&D expenses, Lombardi reported last week. In addition, the company has been spending cash on inventory, capital expenditures, and increasing the headcount of its customer support team, he said, but those expenses are “proportional to the anticipated ramp of the business.”
In the first quarter, for example, Helicos spent about $4 million more than its core cash burn, consisting of about $2.6 million to increase its inventory of HeliScope sequencers, and about $1.5 million for building manufacturing and scientific informatics infrastructures.
“We carefully model both the level of expense and the category of expense to assure that they are consistent with the needs of the business.”
Projecting the core cash burn to the end of the year would leave the company with $5.5 million in cash. However, Lombardi mentioned several ways in which the company could increase its cash position this year.
First, it could borrow the $10 million that remains in a $20 million credit facility from GE Healthcare Financial Services (see In Sequence 1/8/2008). But that withdrawal is “subject to conditions that we must satisfy prior to any borrowing of the second tranche,” he said without elaborating.
Also, he pointed out that the company already expensed the inventory cost for six HeliScope sequencers last year, which included the unit shipped to Expression Analysis. Thus, if it sold five more instruments this year and recognized revenues from all six units, this would provide the company with an additional $8 million to $9 million, he said. However, Lombardi stressed that this does not mean the company plans to ship six systems this year.
Finally, the company might look for additional funding. “We carefully model both the level of expense and the category of expense to assure that they are consistent with the needs of the business, and we will consider additional financing and be opportunistic in pursuing alternatives,” Lombardi said.
Helicos currently has an inventory of 12 instruments, not all of which are assembled, which is “sufficient to meet our initial ramp of the business,” according to Lombardi. Five instruments in manufacturing are undergoing final testing, two for the company’s own R&D efforts and three for future customers. “Our big focus is now to take the knowledge we have gained from building [these instruments], create a better process … and we are focusing a lot of our efforts now on how to best streamline the testing,” he said.
More Apps, Better Tech
On the technology side, Helicos is now using a new version of its sequencing chemistry that has a lower error rate than the previous one and allows it to produce high-quality data in a single sequencing pass. Previously, it had to sequence each DNA strand twice for sequencing applications in order to lower the overall error.
Though the new error rate is “still around 5 percent,” Lombardi said, “what is really important is, even with that high an error rate, the part of the error budget that constitutes substitution errors is still very low.”
At the Cambridge Healthtech Institute Next-Generation Sequencing meeting in San Diego last month, Helicos Chief Scientific Officer Patrice Milos showed data for one-pass as well as two-pass resequencing of the p53 gene for SNP detection.
Company researchers have sequenced a long-range PCR product of the p53 gene that covers exons 2 to 11 in five HapMap samples, she reported, as well as additional upstream PCR products, which she did not discuss.
A large part of the p53 gene region contains repetitive sequence and is AT-rich, she said, making it “a good fragment to practice SNP detection on.”
The known SNPs from the HapMap database provided “a landmark to determine how accurate we are with respect to HapMap SNPs in our algorithm development,” she said.
Both one-pass and two-pass sequencing allowed the researchers to detect known SNPs, as well as to discover novel SNPs that were not contained in the HapMap database.
Helicos was able to identify several SNPs that Sanger sequencing missed because the latter was unable to cover the entire gene region. Sanger sequencing called one SNP incorrectly as a heterozygote that Helicos detected correctly, Milos noted, and Helicos incorrectly called a homozygous SNP in a homopolymer stretch that Sanger identified as a heterozygous SNP.
Helicos is currently assessing the depth of sequencing needed for one-pass SNP detection and is improving its internal SNP calling algorithms. In addition, the company is evaluating other SNP detection software, for example PolyBayes from Gabor Marth’s group at Boston College, according to Milos.
The company is also now sequencing mixtures of DNA samples “to build experience with SNP detection in tumor samples,” she said.
The p53 study is “really just a demonstration of a small target region,” she said, and Helicos is now also looking into sequencing several megabases of human target sequence for SNP detection.
Lombardi said during the conference call that Helicos’ effort to develop a multiplex sample prep process for selecting candidate regions from the genome for sequencing is “still in early stages of development.” The company is establishing collaborations in this area, he said, and plans to publish the method and to commercialize it later this year.
A second focus of Helicos’ research is to develop an amplification-free digital gene-expression assay, Lombardi said.
That application, he said, is of special interest to potential pharmaceutical customers. In April, he mentioned, Helicos and Expression Analysis co-hosted a meeting for large pharmaceutical customers in North Carolina, which “virtually all the big pharmas in North America” attended.
“While interest is high across the spectrum of application areas, it is notable in digital gene expression,” Lombardi said. “Pharma is interested in asking new questions of the transcriptome to better understand disease pathways and pharmacogenomics in general. This interest was a major driver for Expression Analysis in their selection of the Helicos platform.”
Milos mentioned during her talk at the San Diego conference that Helicos has already analyzed the yeast transcriptome by tag sequencing from the 5’-end of transcripts, using an amplification-free 5’-mRNA preparation method. The researchers found close to 6,000 transcripts in the experiment, she said.
Helicos plans to announce additional collaborations in the area of digital gene-expression analysis over the next several months and to publish its initial results, Lombardi said. The company wants to launch product specifications in the second half of the year.
During her talk, Milos also showed proof-of-concept data for a one-pass paired-end sequencing strategy, using the same p53 region that the company sequenced in its SNP detection project. The paired reads covered the entire p53 region, she said.
The scientists hybridized p53 DNA fragments 150 to 170 nucleotides in length to the flow cell surface and sequenced the first 25 bases. They then performed two rounds of “controlled dark fill,” during which they added three types of unlabeled bases at a time to generate “dark” inserts of about 50 bases per round. They then resumed sequencing for approximately another 25 bases using labeled nucleotides.
Helicos is now working on controlling the length of the “dark fill” to ensure evenly sized spacers. It also wants to improve the current 25-percent yield of those strands that contain both read pairs, Milos said. Finally, company researchers are working on generating paired-end reads from longer fragments, up to 10 kilobases in size.
“We know that we can put DNA molecules as large as 10 kilobases on the surface, we know we can read one end of these molecules, we now have to fill and then determine how far up the molecule we can read into the DNA that we are interested in,” Milos said. The goal is to read the ends of strands that are 2 kb, 5 kb, and 10 kb in length, in addition to shorter fragments.
In principle, it would be possible to repeat the dark fill-sequence cycle more than once, she said in response to a question from the audience. “The only challenge there is yield, how far you can go with each addition [of nucleotides].”
Milos also commented on other research at Helicos. She said the company is now routinely sequencing BACs and is moving to bacterial genomes and will “continue to walk up the evolutionary ladder of organisms to get to the point where we are sequencing whole human genomes.”
Helicos is also “very interested” in getting into ChIP sequencing and methylation studies, she said, applications that are “new areas for us over the coming year.”
It is also exploring diagnostic applications of its technology, for example through its collaboration on cancer gene resequencing with researchers at the City of Hope, which it announced last month (see In Sequence 4/15/2008).
Regarding its technology, a major focus over the next few months will be to understand why some reads terminate after less than 20 base additions, while others keep going for much longer. This would enable the company to increase its average read length of approximately 25 base pairs.
“We know we have strands that extend well beyond 50 to 60 nucleotides,” Milos said. “The problem is, they are not the majority. So the question is, how can we convert those strands that are 18, 19, or 20 nucleotides to be more like the 50-nucleotide strands?”