Gene Genies

Jay Flatley, Illumina, and Doug Amorese, Agilent, go head-to-head on one of pharma's hottest topics.

NGP. Genomics and proteomics have been described as the future of the life sciences/pharmaceutical industry. In what areas do you see the most potential?

JF. Genomics studies – including those that analyse genetic variation and function and those exploring the emerging field of epigenetics, or regulatory-driven changes that affect gene expression, amino acid production and protein expression, will enable life scientists to better understand the genetic basis of disease and key pathways for disease progression. With this better understanding comes a better ability to understand disease predisposition, accurately diagnose disease, develop more effective therapeutic approaches, intervene at earlier-stage symptomatic development, and improve clinical outcomes.

As we develop the ability and information to deploy true pharmacogenomics (PGx) strategies and more personalized clinical approaches, benefits should accrue to virtually all participants in the healthcare value chain, including insurance companies, providers, physicians, pharmaceutical and biotech firms, and, most importantly, patients. We’ll encounter a few bumps on the road, such as the slow death of the blockbuster drug model, data privacy and reimbursement challenges to accommodate non-standardized dosages, but it’s all part of an outcome-driven approach that is increasingly enabled by new information and discovery activity.

DA. I do expect genomics and proteomics to play increasingly significant roles in life science research and product development and testing in both pharmaceutical and diagnostics companies. There are three factors that will drive this: the maturation of measurement tools, the development of informatics tools that enable data integration, and the recognition that a question can be answered more completely by measuring or observing it from multiple perspectives. This latter point has never been the gating item. Independent of whether a researcher had training in nucleic acid or protein biochemistry, they’ve known all along that they could get a better understanding of a problem or a biological process by understanding the discoveries coming out of the other disciplines. The challenges have really been in getting high quality data that offers a ‘complete’ view of what is happening within a cell and being able to correlate the information coming from one perspective with information coming from an orthogonal perspective.

I think perhaps the greatest potential is in the area of personalized medicine, where more detailed diagnosis is coupled to specific therapeutics. Cancer treatment will be one of the first areas to take advantage of these advances, in which particular case, it is actually ‘personalized’ to the tumor type. We’re already seeing results from several studies focused on using genomics to type or further characterize tumors.

NGP. In what ways are your own technologies supporting next generation research in this field?

JF. Illumina has developed a broad portfolio of integrated array-based solutions that deliver industry-leading sample throughput, data quality and cost. As a result, researchers have been able to expand experimental scale and quickly generate high-confidence answers to a growing set of biological questions targeted at understanding the genetic basis of disease.

Most common diseases are multi-factorial in nature – involving multiple genes on multiple chromosomes. Of course, they’re aided and abetted by diet and environmental factors. We need to explore disease etiology on a genome-wide basis to help establish root cause and provide a path for enlightened therapeutic development and treatment.

A good example is the exploding field of disease association studies, made possible by the treasure trove of data generated in the just-completed HapMap Project. Illumina is now deploying ‘TagSNP’ content on microarrays that are ‘driven’ by our very powerful Infinium assay. Infinium solutions are enabling researchers to query genetic variation across the entire genome with just 250,000–500,000 SNPs instead of the entire population of 10 million. We’re also delivering quite remarkable data quality – customers are reporting call rates of over 99 percent with reproducibility approaching 100 percent. What this means is that the customer can analyze fewer SNPs than with competing approaches, but achieve the same statistical power for disease association. Investigators can, therefore, place a lighter load on their informatics groups and get answers faster.

DA. The flexibility of our microarray manufacturing process has enabled both us and our users to easily and quickly explore many of the newer genomic research fields. Customers have always had the opportunity to obtain custom arrays from Agilent, but now our eArray website makes the process even easier. This site allows customers to design arrays using probes previously selected by Agilent in addition to those designed in their own laboratories. This flexibility, coupled with our high quality DNA synthesis, has enabled researchers to use our arrays for aCGH, ChIP-on-chip and microRNA analysis.

NGP. How are your services addressing the industry’s need for faster, more cost-effective studies? Are there other demands?

JF. It’s somewhat artificial to separate services from products in as much as both models leverage the same array and assay technologies. Where capital is constrained or where research programs are individually funded and not part of an ongoing program, the service model provides obvious advantages. The service model can also benefit disease consortia, who want to combine samples and increase the statistical power of results without dealing with build-out of a new infrastructure.

That said, the advent of truly disruptive technologies, like those developed by Illumina, will deliver the same experimental benefits to those purchasing a service as to those investing in a turnkey system for genetic analysis. If you do the math, the capital cost for large-scale biological experimentation is somewhere around a 10 thousandth of a cent per data point. Arrays, reagents and labor are much bigger cost drivers.

DA. We view achieving a lower cost per experiment to be a natural progression of any technology that is going to truly add value and survive. In our case, flexibility was a driver in our decisions, leading to the development of our array manufacturing system. We next focused on achieving the optimum DNA density within a feature and synthesizing the highest quality DNA possible. We have synthesized oligo nucleotides up to 150 bases in length for some of our collaborators. We’re now turning our attention to feature density. While data quality clearly has an impact on the cost of a study, this can be overlooked. There have been tremendous advances in inkjet printing over the past few years. Our next generation arrays take advantage of these advances. We have been providing early access 185k arrays to a few of our customers and expect to offer these arrays to others as our internal capacity increasing with full commercial release early next year. These high-density arrays enable multiplex of sample processing as well. Currently we are offering eight sub-arrays on a single 1x3 slide. As customer needs change, we have the flexibility in our system to move with those changes. Going back to one of my earlier points, neither higher density nor lower cost are of value if data quality goes down so every change that we make needs to take into consideration overall system performance. While I’m certain our customer value the flexibility we offer, I believe most of our customers have selected the Agilent platform because it offers the best sensitivity and specificity. So we cannot let any of our changes compromise that.

NGP. I understand you also offer genome-wide expression analysis. What does that entail and why is it so beneficial?

JF. Genome-wide expression analysis provides a measure of RNA abundance, which can help serve as a proxy for protein expression, which is currently impossible to measure on a ‘global’ basis. Our tools provide the ability to profile gene expression to determine what genes are turned on and off, or up-regulated or down-regulated in diseased vs. normal tissue. Data analysis helps us identify genes – and associated proteins that might represent drug targets – that might be implicated in a disease pathway or other biological pathways.

Increasingly, expression profiles are being correlated with more standard diagnostic tests, in as much as they can provide a molecular signature of a disease phenotype and aid with earlier diagnosis.

DA. It is often important to view essentially all of the changes taking place within the cell as it relates to transcription to get an accurate view of how the cell is responding or what constitutes the ‘normal’ state within a cell. Other times, researchers have already identified a pathway or set of pathways that they want to study in greater detail. Our customers can select a pre-defined complete genome probe set or do their own probe design to detect all of the transcripts that may be present within a cell.

Alternatively, they do a little of each. If a customer chooses to use the probes we’ve designed, they have the additional benefit of knowing that they were empirically validated. The flexibility of our system has enabled us to synthesize many probe candidates and then select the best performing probes from multiple candidates through differential hybridization experiments.

NGP. A growing range of arrays now exist for gene profiling. Can you tell me more about what’s available? What are the key characteristics that customers demand from these arrays?

JF. If by profiling, you mean gene expression profiling, Illumina offers a growing list of expression arrays that can generate RNA profiles of whole genomes, focused sets of gene content, and custom content.

Illumina is the only firm to offer a microarray that can provide whole-genome RNA profiling of six samples (48,000 transcripts apiece) on the same device. In gene expression, like genotyping, we’ve improved performance and driven price points to breakthrough levels that enable researchers to process more samples and secure better statistics at the same price.

Data quality is critical. Our Sentrix Arrays have high sensitivity to low-abundance transcripts and high array-to-array reproducibility. One of our customers, who studies embryonic stems cells, has found that many genes determined to be 2X or more up-regulated are, in fact, false positives, meaning that scientists have been chasing too many fictional targets. A more enlightened approach would be to identify transcripts that are significantly, or reproducibly, over-expressed from sample to sample. Thus a target that is up-regulated at just 1.2 or 1.3 times normal tissue, but so regulated on a significant basis, can represent a highly significant potential target and not be excluded from further analysis simply because it doesn’t comply with the traditional 2X threshold.

Another issue for customers has been the challenge of data comparability between platforms. There are currently several important studies in process that will enable users to invest in higher-performance platforms and still correlate new experimental data with libraries of older data generated on a competing platform. Cross-platform comparability initiatives will also help researchers ascribe experimental variability to its proper source, whether it’s platform-based or driven by lab-to-lab differences.

DA. We offer arrays with empirically validated content as previously described for human, rat and mouse. From these experiments, we’ve refined our probe selection process to better ensure our ability to select good probes to represent essentially nucleic acid sequence. We’ve made these tools available to researchers studying a variety of model organisms to help them develop whole genome arrays for these species. These collaboratively designed arrays are available to all of our customers through our eArray website.

The requirements of most of our human, rat and mouse microarray customers are quite straightforward; in essence, they want a microarray that faithfully represents the genome with the highest sensitivity possible. For those studying model organisms, it is essentially the same with the additional desire to update and or change probes at will.

NGP. Looking forward, how positive are you that the potential of genomics will be realized and what will have to happen to do so?

JF. Disease association studies will grow rapidly as scientists begin to leverage HapMap data to understand the genetic basis of disease. Furthermore, microarrays are already migrating from the lab to the clinic – proof that the methods are delivering good information, reproducibly, at the right price points. And there’s significant downstream promise in the field of pharmacogenomics (PGx) – delivering the right drug to the right person at the right dosage level. So we think we’ve just begun to tap the potential of genomics.

DA. I’m very confident that genomics will play an increasingly important role in pharmaceutical product development and disease characterization. My optimism is at least in part due to the advances that have been made in the past few years by researchers everywhere that are using these tools and the direction I see our technology moving. The tools have reached a level of maturity that they can be counted on for accurate data, and informatics packages have been created that enable researchers to get useful information from this data and to integrate this information with information coming from different disciplines or different prospectives.

What is necessary now is for customers to decide on what information is most relevant to them for a specific question, to customize these tools to provide the relevant information at an affordable price and to integrate that information in a meaningful way to better understand a biological process, the mode of action of a lead compound, etc. Fishing can be a lot of fun and a lot can be learned in the process, but for the full potential of the technology to be achieved it needs to be targeted.