Affymetrix Blog

Top 6 trends in translational science in 2012

Frank Witney — Fri, 18 Jan 2013 17:56:26 +0000

It is that time of year when we reflect on the past year and embark on goals for the new one. In keeping with this tradition, the aim of this first post of the New Year is to reflect on the major discoveries in translational science achieved in 2012 that I believe will shape 2013 and many more years to come.

In no particular order, I believe these are the top five trends in translational science this year:

The ENCODE project

One of the most intriguing questions in genomics has been whether the majority of the DNA in the genome serves a particular function. Sequencing of the human genome had previously revealed that only 2% accounts for protein coding genes and hinted that the remaining DNA, often called “junk DNA,” had no apparent (or at least identified) function – a fact that has puzzled scientists for decades.

Recently, the successful completion of an international, multi-center, and multi-disciplinary project called ENCODE (The Encyclopedia of DNA Elements) has provided an answer to this question. Almost 80% of the “junk DNA” that was previously thought to have no function actually contains switches that determine when and where a gene may be activated and expressed. A detailed map of 4 million switches, published in 30 papers in the journals Nature, Genome Biology, and Genome Research, now provides scientists with more insights into which switches are involved in the activation of genes.

The ENCODE project is the trigger for scientists to consider gene expression not as the result of linear regulatory elements that reside close to the genes, but as the result of a dynamic network of switches dispersed all over the genome.

Breast cancer is not just one disease

A paper that appeared in Nature in April, 2012¹, challenged our current perception of breast cancer. An elegant high-throughput study performed by researchers around the world defined breast cancer as a group of ten different subtypes, each one characterized by a particular number of aberrations that show distinct gene expression architecture. Essentially, the study proved that breast cancer is not just one disease, but ten different ones.

This study has profound research and clinical implications: scientists will now need to look into ten different diseases, but they can use this information to create therapeutic interventions and diagnostics targeted to a specific disease sub-classification. Hopefully this will aid development of new generations of personalized medications that will improve clinical outcome and prognosis for breast cancer sufferers worldwide.

Cell-to-cell variation

Traditional approaches to studying genomics and transcriptomics of a given sample have been based on a group of cells (i.e., a given tissue that is composed of multiple populations of cells). Consequently, results of these analyses are representative of multiple numbers of cells. However, advancements and improvements in techniques have allowed scientists to decrease the amount of starting material, and most recently, they have even permitted the study of single cells. Single-cell analysis is helping the detection of minute differences in the population of cells. This information on the individual responses of cells can give additional insight for enhanced drug design and diagnostic methods. The tools for routine single-cell analysis are still being developed and refined, but scientists are already looking at the activity of single cells to gain a better understanding of cell function with respect to the micro-environment.

Circulating tumor cells

In the past few years, circulating tumor cells have attracted the attention of cancer researchers worldwide. This population of cells originates from clones of the primary tumor. As the tumor develops, they leave the primary location and begin circulating in the bloodstream. Because their molecular characteristics are similar to those of the primary tumor, they can be used to aid diagnosis and provide insight into disease progression and clinical management, especially when the primary tumor is not easily accessible.

In 2012, we have witnessed tremendous progress in the development of assays and techniques to identify and characterize these cells so that we can study them in greater detail and realize potential clinical benefit. To support these efforts, Affymetrix sponsored a webinar with three world-renowned experts in the detection and characterization of circulating tumor cells; you can view this webinar by clicking here.

Microarrays supplant karyotyping

Karyotyping is one of the most prevalent cytogenetic techniques used in perinatal or cancer genetics to identify pathogenic chromosome variations, typically copy number variations. The introduction of microarray-based cytogenetics is enabling scientists to query the entire genome for chromosome aberrations in a more systematic, precise, accurate, and sensitive way than microscopic examination of chromosomes by karyotyping. For example, two recent publications in the New England Journal of Medicine demonstrated how microarray-based cytogenetics is superior to karyotyping in detecting clinically significant findings in embryos and stillbirths^2,3. This is in addition to the well-established use of microarrays in postnatal cytogenetics.

Today, microarray-based methods have a diagnostic yield in postnatal testing of 20% compared to 3% for karyotyping, and this had led the scientific community to establish microarrays as the first-line tool in constitutional genetics⁴. The growing number of publications suggest that cytogenetic analysis will move in the same direction in oncology.

FFPE samples

Formalin-fixed paraffin-embedded (FFPE) samples have lately emerged as a viable alternative to studying frozen fresh tissues for gene expression (both global and small subsets of RNAs), cytogenetic analyses, and biomarker identification. FFPE samples are the most common samples found in histopathology labs worldwide and are routinely made to diagnose and catalog solid tumors after tissue biopsies. The development of appropriate methods for the extraction and processing of degraded DNA and RNA from FFPE samples has allowed scientists to run high-throughput genomic and transcriptomic analyses and most interestingly link this information to clinical outcomes, which are readily available for archived samples. Recent work has also established the ability to detect low copy number RNAs in situ in FFPE tissue samples. Taken together, improvement in FFPE methods in the past year is another confirmation that this difficult sample type will play an increasingly important role in cancer research and diagnostics in the future.

Frank Witney

¹Curtis C., et al.The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups. Nature 486(7403):346-352 (2012).

²Wapner R. J., et al. Chromosomal Microarray versus Karyotyping for Prenatal Diagnosis. New England Journal of Medicine 367:2175-2184 (2012).

³Reddy U. M., et al. Karyotype versus Microarray Testing for Genetic Abnormalities after Stillbirth. New England Journal of Medicine 367:2185-2193 (2012).

⁴Manning M., et al. Array-based technology and recommendations for utilization in medical genetics practice for detection of chromosomal abnormalities. Genetic Medicine 12(11):742-745 (2010).

Translating Genetic Biomarkers to the Clinic: The Promise and Pitfalls of Developing Robust, Reliable Signatures

Frank Witney — Thu, 01 Nov 2012 20:45:05 +0000

The recent Affymetrix sponsored Science/AAAS webinar entitled “Translating Genetic Biomarkers to the Clinic: The Promise and Pitfalls of Developing Robust, Reliable Signatures” [view webinar], highlighted some very interesting aspects of translating biomarker research into products that are ready for clinical use. The Q&A session that ensued included many questions on the topic of creating and validating biomarkers and then translating them into clinical products. Here are some questions from the audience that the speakers did not have time to answer for which I’ve provided some perspectives.

I’d welcome hearing from you!

Frank Witney

What do you see as the future for array-based diagnostics in the era of next-generation sequencing (NGS)?

Recently, NGS methods, e.g., RNASeq, have shown great potential as a general discovery tool and to some extent as a research-based biomarker identification and validation tool. However, members of the scientific community are debating whether, or in what timeframe, NGS can become a viable (cost, accuracy, ease of use) technique in the clinic.

Before this question can be answered, a more important question needs to be asked: what does the scientist wish to accomplish? If the biomarker signature is to be included in a commercial test in a clinical setting, where approval from the FDA is needed, then currently only array-based methods have received this approval. NGS approaches have not yet been validated.

Increasingly other considerations are influencing method selection, for example, sample type and cost considerations. Array-based technologies can analyze samples even in nanogram quantities and can process fresh and formalin-fixed paraffin-embedded (FFPE) samples, whereas NGS approaches typically require a larger amount of sample and often need fresh samples. Although the cost of NGS (price per sample run) is falling, it’s important to also factor in the cost (and sometimes the time needed) to complete the complex bioinformatic analyses of the data. This last factor may result in NGS methods being more costly overall than array-based methods and impractical from a clinical expertise perspective in analyzing the data for clinical outcomes.

Is developing biomarkers for rare diseases an advantage or disadvantage, and why?

Rare or orphan diseases are those that affect only a small percentage of the total population. In the US, they are defined as diseases that affect less than 200,000 people, whereas in Europe, rare diseases are classified as those affecting fewer than 1 in 2,000 people.

In Europe and the US, specific legislation has been introduced to motivate biotech and pharmaceutical companies to develop drugs for rare diseases. These orphan drugs, which, until recently, have been perceived as having limited commercial opportunity, are urgently needed to make a difference to patients, who may have limited or no treatment options.

Drug development for rare diseases can be very challenging, as the diseases usually have a poor prognosis, and primary outcomes for clinical trials can be difficult to measure. Lately, in an effort to accelerate drug research, scientists have suggested that surrogate or biomarker endpoints could be used as clinical study endpoints in orphan drug clinical trials. Therefore, developing biomarkers for rare diseases would be very beneficial for those involved in drug research on rare diseases, and it would ultimately benefit patients.

Do you see any role for variations in non-coding regions as markers for altered gene expression?

A recent multicenter study called Encyclopedia of DNA Elements (ENCODE), published in the journals Nature, Genome Research and Genome Biology, showed that the vast majority of the non-protein coding DNA in the human genome has at least some biochemical function.

This was the first conclusive proof that the so-called “junk DNA” can, in fact, have specific functions, and that a lot of regions may be responsible for controlling the expression of the gene coding elements of DNA. We have only started scratching the surface in trying to understand how these non-protein coding regions may influence the expression of genes. Therefore, variations in non-coding regions may become very useful markers for monitoring altered gene expression in the future.

Should validation be on the discovery platform or a different more specialized subgenomic platform?

After creating your biomarker signature, it is important that the validation of your signature is implemented on the same platform used for discovery but on new patient samples.

Is the medical community ready to interpret these new molecular vital signs in their medical bag?

Frank Witney — Thu, 27 Sep 2012 22:00:32 +0000

A patient visits his/her physician for an annual check-up. The physician takes the patient’s pulse and other vital signs as well as traditional blood chemistries and then refers to the patient’s complete genomic, transcriptomic, proteomic, and metabolomic profile of the past year, which was built by collecting regular blood samples. On the basis of this routine examination and interpretation of the "omics" data, the physician recommends that the patient watch his/her diet and fatty acid intake, as the data revealed an elevated risk of coronary thrombosis. Tests are also ordered to investigate the potential onset of type-2 diabetes. By comparing the current "omics" profile with the previous year’s, the physician concludes that the patient’s health has deteriorated slightly overall, but this can easily be reversed by making simple lifestyle changes.

Such a story is far off in the future — or is it?

A paper published in Cell by Chen, et al.¹ demonstrated exactly this principle — that an integrative personal "omics" profile (iPOP) can be used for both health monitoring and disease diagnosis. This study monitored a healthy 54-year-old male over a 14-month period and analyzed genomic, transcriptomic, proteomic, metabolomics, and autoantibody information from his blood plasma and serum. The aim of the iPOP was three-fold: i) to evaluate disease risk by analyzing an individual’s genome; ii) to study the expression of personal variants; and iii) to use the "omics" data to gain a better understanding of the transition between physiological states. Because the volunteer contracted two respiratory infections over this period, regular sampling at the time of infection also allowed comparisons between healthy and diseased states. The study was very revealing. Not only did it demonstrate that molecular information can be used to estimate disease risk and for monitoring so that the disease can be treated, but it also painted a better picture of the dynamic processes occurring during the transition from healthy to diseased state.

This elegant study demanded the scientific know-how and data analysis skills of its 41 authors. If such profiles and a "molecular medical bag" are to be used for patient care in the future, significant leaps in sample processing and data analysis for clinical utility will be required.

At present, physicians rely on data obtained from monitoring vital signs and biochemical tests, and they have only just started incorporating molecular tests into disease diagnosis and prognosis. In the future, if molecular diagnostics become commonplace, physicians will either need to be trained on how to analyze and translate this information to being clinically actionable, or alternatively, they may have to rely heavily on a much broader scientific community — statisticians, geneticists, genomics specialists, cytogeneticists — who will process and then interpret these new vital signs.

Genomics medical initiatives and new medical residency training programs have begun to take root. However, there needs to be far more funding, emphasis on medical training, and collaboration among scientists in translational medicine, drug discovery — companion diagnostics, and the medical community, in order to gain a deeper understanding of the "omics" data and their clinical utility.

Frank Witney

¹Chen R., et al. Personal omics profiling reveals dynamic molecular and medical phenotypes. Cell 148(6):1293-1307 (2012).

Digging into the archives – unlocking the full potential of tumor samples

Frank Witney — Thu, 30 Aug 2012 23:08:22 +0000

Important reference collections of solid tumor samples lie fragmented in the histopathology labs of hospitals and medical research centers. The reference collections in question consist of FFPE (Formalin-Fixed Paraffin-Embedded) samples. This standard histopathology technique is used worldwide to preserve and catalog solid tumor samples after tissue biopsies and allows samples to be routinely used in diagnostic tests, which is not always possible with other commonly used tissue preservation techniques such as freezing.

Until recently, FFPE samples have not been used to their full potential. They have been predominantly used in methods that examine a handful of genes and only tend to be suitable when there is prior knowledge of the gene target (for example, in histopathology in situ hybridization tests or low-plex gene expression studies). But recent developments in technology have suggested that this archived collection could be a rich and valuable resource for global gene expression, cytogenetic and somatic mutation exploratory studies, and biomarker identification.

The breakthrough that allowed scientists to use FFPE samples for large-scale genomic analyses is a result of advances in sample processing, especially improvements in DNA and RNA extraction methods. Formalin fixation introduces chemical cross-links between DNA, RNA, and protein, resulting in nucleotide fragmentation and short nucleotide segments. Specifically designed nucleotide extraction kits can reverse these cross-links and allow DNA or RNA extraction, which can be used for high-throughput microarray studies after special amplification. Comparative studies between FFPE microarray and frozen tissue samples have shown consistent and comparable results.

The wide availability of FFPE samples, some of which are more than 10 years old, for high-throughput microarray studies holds enormous potential for researchers interested in cancer biomarkers. Those samples can be linked to in situ analyses, histopathology markers, and, more importantly, to clinical phenotypes and outcomes. Therefore, analyzing those “old” samples allows researchers to retrospectively correlate genomic information with clinical outcomes and gain new insights into biomarkers and clinical progression.

This brings us to an important question: Given that a hospital or medical center has a limited budget, which cancer samples should be prioritized for biomarker identification? Fresh samples that allow for robust genomic and transcriptomic profiling or old samples that link genomic information to histopathology markers and clinical outcomes? Should any project for biomarker identification include an approach that uses both sample types? FFPE samples may already contain a wealth of information that awaits discovery, and we have only just to construct the tools that allow us to dig deep into the archives and extract it.

Frank Witney

Genome-wide association studies: which method to choose?

Frank Witney — Wed, 25 Jul 2012 20:14:30 +0000

Genome-wide association studies (GWAS) that identify and correlate single nucleotide polymorphisms (SNPs) to complex diseases are predominantly carried out with SNP microarrays specifically designed to interrogate millions of different polymorphisms in the human genome (and in the genomes of other organisms). The results are then typically cross-referenced with data from the HapMap Project or the 1000 Genomes Project in a process called imputation that aims to substitute values for missing data. Since making their debut in 2005, GWAS have contributed to numerous published¹ studies and a better understanding of diseases.

Recently, an alternative method for conducting GWAS appeared in the literature and stirred numerous discussions among scientists. In Nature Genetics, Pasaniuc, et al.² described a GWAS method using low-coverage sequencing and imputation using information from the 1000 Genomes Project. Using simulations and experiments, the authors demonstrated that sequencing of libraries at very low coverage (0.1–0.5x) and less frequent variants coupled with imputation can be a viable approach for determining common in GWAS. They compared their results to those obtained by SNP arrays and demonstrated comparable results without many false positives.

It is still too early to judge whether this technique will be a viable alternative to the array-based GWAS, as it has not yet been validated in an actual GWAS. However, in one of my previous posts, I discussed not only the need to choose robust approaches for experiments but the need to validate those results with other appropriate high-throughput (HT) experimental approaches. This is especially true for studies meant to create data that will subsequently be used for further downstream analyses and clinical applications. Scientists need to carefully consider the limitations of the method they are using and to choose the most appropriate approach to provide robust data while keeping in mind that orthogonal techniques are required to validate the data.

The authors of the paper pointed out several caveats in their method. Analyses of sequence-based GWAS are more difficult than SNP arrays because the computational methods are still under development and not as readily available to researchers. In addition, the analysis pipeline of sequencing data is considerably more challenging than the standardized analysis of genotyping studies. More importantly, the method may not be suitable for the discovery of less frequent and rare variants.

In contrast, population-specific SNP arrays have recently allowed scientists to extend GWAS findings from European populations to populations of different ethnic origins [view video: NIH Grant; A Grand Opportunity: Developing a Resource for Genetic Epidemiology Research in Adult Health and Aging]. These SNP arrays include coverage of disease-associated common and rare variants as well as indels and will likely provide a better opportunity for researchers to detect less frequent disease alleles that have meaning to our ethnically diverse populations around the world.

Finally, a large part of the discussion is dedicated to how the sequencing GWAS can be a cost-effective alternative to SNP arrays in GWAS. The authors conducted a thorough survey of the costs and demonstrated that owing to cost reductions in sequencing and handling of genomic libraries, the approach can attain a higher effective sample size than SNP array studies within a set budget of $300,000. I agree that the cost of tools, reagents, and consumables is an important factor that scientists need to consider when deciding which method to use for their HT experiments, but it shouldn’t be the sole determining factor. Instead, the ability to provide solid, validated, and robust results should remain the primary basis for any experimental method.

Frank Witney.

¹Hindorff L. A., et al. Potential etiologic and functional implications of genome-wide association loci for human diseases and traits. Proceedings of the National Academy of Sciences of the United States of America 106(23):9362-9367 (2009).

²Pasaniuc B., et al. Extremely low-coverage sequencing and imputation increases power for genome-wide association studies. Nature Genetics 44(6):631-635 (2012).

What does a united Affymetrix and eBioscience mean for the scientific community?

Frank Witney — Mon, 02 Jul 2012 17:51:54 +0000

I am really excited about the new possibilities that the Affymetrix and eBioscience union enable for translational sciences and beyond. We are working to help researchers integrate genomics, proteomics, and cell biology for the development of personalized medicine. The acquisition of eBioscience will allow us to expand beyond genomics and transcriptomics to encompass cell-based assays and immunoassays. We can provide solutions to analyze biological processes at the molecular, cellular, and pathway level and to empower scientists to move their findings from the lab bench to the clinic.

Imagine a world where:

We can benefit from understanding our own DNA
We can understand disease at the molecular level, detect it earlier, and provide personalized treatment to individuals
We can help prevent disease and focus on our wellness

Advances in genomics, proteomics, and pathway biology have the potential to revolutionize our way of life.

With eBioscience, Affymetrix will provide more comprehensive solutions enabling translational medicine and molecular diagnostics and offer new products in key applications of immunology, oncology, cell biology, stem cell biology, and diagnostics. Ultimately, the joining of eBioscience with Affymetrix is another step toward offering more complete solutions for personalized medicine and helping scientists to tell their own story as part of the Genome Generation.

Frank Witney

Translational sciences: Shouldn’t we go beyond genomics?

Frank Witney — Wed, 30 May 2012 21:42:52 +0000

Cancer is one of the most characteristic examples of a disease for which the scientific community has come together to systematically produce vast amounts of genomic data. The aim is to gain a better understanding of this disease in order to create targeted therapeutic compounds and companion diagnostics, and improve cancer care. These large, ambitious projects are not unjustified – cancer is not only one of the most common diseases afflicting the modern world, but also one of the most complex.

At the genomic level, our understanding of cancer points to the existence, or gradual accumulation, of genomic alterations that are responsible for the disease manifestation; these alterations are commonly referred to as "drivers."¹^,² There are, however, innocent bystander genomic alterations that do not have an oncogenic potential; these are commonly referred to as "passengers." Therefore, any attempts to understand how cancer appears, evolves, and spreads may need to start with the fundamentals of separating the "passengers" from the "drivers" and then attempting to create detailed catalogs of genomic alterations in different cancer types. This is the goal of large-scale international collaborative efforts such as The Cancer Genome Atlas (TCGA) and the International Cancer Genome Consortium (ICGC). More recently, approaches to the isolation and characterization of circulating tumor cells³ may provide an additional view to understanding the molecular drivers of cancers.

TCGA is now in the middle of an ambitious five-year program to chart the genomic changes of 20 different types of cancer – brain, breast, gastrointestinal, gynecologic, head and neck, hematologic, skin, thoracic and urologic – by using a variety of high-throughput methods. At the same time, ICGC is attempting to go one step further by aiming to obtain comprehensive catalogs not only of genomic, but also transcriptomic and epigenomic changes in 50 different tumor types.

Those ambitious projects as well as numerous others have generated vast amounts of data that are beginning to unravel the complexity behind cancer. Recent findings suggest the existence of intratumor heterogeneity which can lead to underestimation of the complexity of the tumor genomics landscape.⁴ Molecular subclassifications of tumors are being characterized and found to have different combinations of driver mutations.⁵

So how does understanding the molecular etiology of cancer translate to personalized medicine and improving clinical outcomes? Earlier studies have led to the generation of new therapeutics. For example, it was almost 30 years ago that the RAS gene family was implicated in cancer, but its usefulness as a biomarker in cancer management has only recently been realized. In contrast, thanks to current genomic approaches, it took approximately 10 years from the time that the BRAF V600E mutation was identified, to the market approval of a drug that targets this mutation in melanoma cancer patients.

Our current understanding of cancer, however, is far from a picture painted by the sum of genomic, transcriptomic, and epigenomic events. Though these are important, scientists know that cancers need to be studied in the context of their environment and developmental stage. Ultimately, scientists need to understand how findings at the molecular level translate in the patient and give insights on an epidemiological level – going beyond genomics to understanding the underlying cellular mechanisms.

Exciting times lie ahead for cancer researchers. While they face significant challenges in being able to interpret data from high-throughput experiments and in processing, synthesizing, and making sense of this information, once they have this at their fingertips, they will be able to drill down and find answers to the scientific questions that really matter – those that can be translated to improving clinical outcomes.

Frank Witney

¹Chin L., et al. Cancer genomics: from discovery science to personalized medicine. Nature Medicine 17(3):297-303 (2011).
²Chin L., et al. Making sense of cancer genomic data. Genes & Development 25(6):534-555 (2011).
³Garnett M., et al. Systematic identification of genomic markers of drug sensitivity in cancer cells. Nature 483(7391):570-575 (2012).
⁴Gerlinger M., et al. Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. New England Journal of Medicine 366(10):883-892 (2012).
⁵Curtis C., et al. The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups. Nature Epub (2012).

Validomics: How do we ensure biologically relevant data?

Frank Witney — Mon, 14 May 2012 16:06:00 +0000

In my previous post, I talked about the wealth of information being produced by high-throughput (HT) genomics methods and how post-genomic research is helping to paint a new picture of living processes. It was only very briefly that I touched upon and stressed the importance of data validation, particularly using additional experiments and orthogonal methods to validate results from the HT approaches. I could not have predicted what was about to appear!

Six days later, an article appeared in Nature pointing out the dangers of misinterpreting data from HT RNA-seq experiments¹. While the topic is certainly not new, the article in question mentioned two recent high-profile publications involving RNA sequencing on imprinted genes² and RNA editing³. In both cases, it is alleged that errors or misinterpretations in the data analyses led the researchers to overestimate their findings, which were subsequently challenged by other researchers, casting yet another suspicious eye on the field of HT genomics.

While it might be unfair to shine the spotlight only on these individual publications, the situation begs the question: to what extent is misinterpretation of data happening in HT experiments in general and thereby finding its way into publications? Even worse, how many other scientists are basing their assumptions on wrong data or insufficiently analyzed results⁴?

In many other areas of science, orthogonal techniques are used to validate data and improve the confidence of functional analysis. For a long period of time, it has been common practice to validate microarray results by performing real time PCR experiments on a handful of "standard" genes. Why are we not applying the same principles to data generated with next-generation sequencing (NGS) technologies?

Today the research community has a variety of HT methods available to them, and the results of one method can be readily validated by using a different orthogonal method; for example, why are NGS data not validated using genome-wide technologies such as microarrays or multiplex analysis methods such as branch chain DNA technologies? This is beginning to happen, as there is recent evidence in the scientific literature^6-9 where scientists used microarray experiments to validate results of RNA-seq analysis. De novo discovery with NGS techniques is one way to expand the envelope of our knowledge; however, we must verify that we are all using biologically-relevant data. This is true for newly discovered RNA or DNA variants.

With the looming appeal of the $1000 genome, as well as other newer technologies, we must focus on the real "endgame"– reducing false discoveries and driving clinically-actionable discoveries. As Dr. Atul Butte from Stanford School of Medicine recently said, "…As scientists we have to share our data… We're in a data driven revolution. Data is power." When the future of translational research and translational medicine/personalized medicine is dependent on integrating the volumes of genomic, clinical, and patient data, the scientific community must ensure that all these data are validated, irrespective of the additional time and resource constraints.

If scientists are to stand by their results, as the principal investigators of the publication mentioned in the Nature article, how can the silver lining be found? It's through validomics using well-characterized orthogonal methods. Should the more germane question be "Can there ever be enough validation?" or should it be "Is there enough?"

Frank Witney

¹Hayden E. C. RNA studies under fire. Nature 484(7395):428 (2012).
²Gregg C., et al. High-resolution analysis of parent-of-origin allelic expression in the mouse brain. Science 329(5992):643-8 (2010).
³Deveale B., et al. Critical Evaluation of Imprinted Gene Expression by RNA-Seq: A New Perspective. PLoS Genetics 8(3):e1002600 (2012).
⁴Git A. Research tools: A recipe for disaster. Nature 484(7395):439-40 (2012)*
⁵Piston D. W. Research tools: Understand How it Works. Nature 484(7395):440-1 (2012)*
⁶Carter S. L., et al. Absolute quantification of somatic DNA alterations in human cancer. Nature Biotechnology. Epub (2012).
⁷Iacobucci I., et al. Application of the whole-transcriptome shotgun sequencing approach to the study of Philadelphia-positive acute lymphoblastic leukemia. Blood Cancer Journal Epub (2012).
8Hackett N. R., et al. RNA-Seq quantification of the human small airway epithelium transcriptome. BMC Genomics 13:82 (2012).
9Bradford J. R., et al. A comparison of massively parallel nucleotide sequencing with oligonucleotide microarrays for global transcription profiling. BMC Genomics 11:282 (2010).

*Articles discuss the pitfalls of reaching the wrong conclusions because of limited release of information on product ingredients by reagent companies and ill-understood automated experimental methods, especially by young researchers.

Genomics, transcriptomics, proteomics – is it time for validomics?

Frank Witney — Thu, 19 Apr 2012 22:02:26 +0000

Today we celebrate DNA day, which commemorates the publication of the structure of DNA in 1953¹. On this very day in 2003, the Human Genome Project (HGP) was completed, providing us with a reference sequence of the human genome². The draft sequence had been provided two years earlier, in 2001³^,⁴. Since the completion of the HGP, tremendous progress has been made in the field of genomics, but there is criticism that the anticipated benefits in improved human health have yet to be realized. Some have claimed that that the likely benefits of the HGP were initially exaggerated to obtain funding for the project.

We recognize that knowledge gained from the human genome has resulted in great advances in understanding its structure and function but only a small number of new therapeutic approaches, to date. But while overall it may have not yet provided the genomic revolution everyone expected, are we being short-sighted and not recognizing that the genomic revolution is, in fact, happening now?

Last year, for the 10-year anniversary of the draft sequence, the scientific community behind the achievement published a paper⁵ presenting an updated vision of the post-genomic era. In the paper, they reiterated that the human genome sequencing was just the beginning of a journey and not the end. In their updated genomics vision, the scientists outlined how the road to improved healthcare will be achieved through five distinctive steps: 1) understanding the structure of genomes 2) understanding the biology of genomes 3) understanding the biology of disease 4) advancing the science of medicine and, ultimately, 5) improving the science of healthcare.

We are currently in an era where we are starting to gain a better understanding of the biology of genomes and disease. The sequencing of the human genome has provided comprehensive resources for genomic data (which we will speak about in another blog post). In addition, the numerous "omics" fields that have appeared (transcriptomics, proteomics, and metabolomics, as well as many others) and systems biology are providing us with new ways to study living processes. Finally, the fundamental unit of the body is the cell, and it will be crucial for new methods to evolve that allow measurement of molecular events in single cells.

Isn’t it also time for "validomics," where scientists systematically validate the data and link them to phenotypes or clinical data? Isn’t it time that the scientific community departed from Eurocentric data for genetic analysis and adopted a more ethnic-specific approach to drug making? In order to continue discovering new processes (e.g., ncRNA) and their influence on and/or stratification of disease, don’t we need to ensure that we are building our scientific knowledge on the backbone of validated data?

For Affymetrix, the "Genome Generation" is the name we use for the current genomic revolution of which the whole scientific community and the public are a part. The purpose of the "Genome Generation" blog is to highlight technological and scientific advances of the post-genomic era; to discuss how these advances can help scientists to gain a better understanding of living processes; to discover how this knowledge relates to disease; and ultimately to explore how it can be used to create new drugs and new diagnostics.

We have started to hear from you at the Human Genome meeting (HUGO 2012) in Australia and the American Association for Cancer Research (AACR 2012) meeting in Chicago. The overwhelming responses are diverse, and here are just a few examples:

"…computational approach to studying the molecular mechanisms behind complex diseases…from genotype to a candidate disease gene"

"My Mom had stage III CKC cancer…she is cancer-free for 6 years!"

I will visit many of themes in our subsequent posts, but more importantly, I would like to hear from you. Therefore, do not hesitate to leave a comment and start a discussion on a topic you are passionate about.

Frank Witney

¹Watson J. D., Crick F. H. C. A Structure for Deoxyribose Nucleic Acid. Nature 171:737-8 (1953).
²International Human Genome Consortium. Finishing the euchromatic region of the human genome. Nature 431(7011):931-45 (2003).
³Lander E. S., et al. Initial sequencing and analysis of the human genome. Nature 409(6822):860-921 (2001).
⁴Venter J. C., et al. The Sequence of the human genome. Science 291(5507):1304-51 (2001).
⁵Green E.D., et al. Charting a course for genomic medicine from base pairs to bedside. Nature 470(7333):204-13 (2011).

The Genome Generation

Frank Witney — Sat, 10 Mar 2012 00:49:09 +0000

Traditionally, a visit to the physician started with a discussion–ranging from the description of symptoms, to your medical history and family profile–and continued with a review of classical blood chemistry results, a physical examination and sometimes more advanced diagnostic tests such as biopsies and/or imaging. But, lately, two new tools have been added to the doctor’s bag to diagnose and treat diseases: genomics and genetic tests.

The sequencing of the human genome¹^,² and the implementation of functional genomics are enabling scientists and physicians to examine disease at a more fundamental molecular level, often extending beyond the observed phenotype in an attempt to understand how our genes and the environment interact to influence it. For the complex diseases that plague our society (e.g. cancer, cardiovascular, and neurological maladies), molecular medicine offers the hope of sub-classifying diseases, increasing precision of treatment through personalized drug regimens and improving quality of life.

We are now living in a generation best described as the “Genome Generation”³, during which advances in genomics and genetics have the potential to revolutionize the way we practice medicine. The Genome Generation can be described in two key ways: First, there is an unprecedented amount of information from large-scale genomic and genetic studies residing in growing databases, from which the impact of genetics, genetic variants, and gene expression on disease outcomes can be studied. Secondly, it is about how this information can be used in diagnostics and medicine, how it can divide patients with the same disease into subgroups, and how it can influence physicians on the type and course of treatment they recommend to patients. It is about providing powerful diagnostic tests to detect and segment disease and direct the development of safer, more effective medicines.

For example, genomic methods are increasingly used to advise parents in prenatal, IVF and postnatal settings, for early disease detection and classification, to customize drug therapy dosing, to discourage application of expensive drug regimens to non-responders, and to help reduce drug side effects. The Genome Generation has created a community of academic and clinical researchers, molecular tool providers, patients, ethicists and payers with common goals: to improve patient outcomes but also to mitigate spiraling health care costs in light of our globally aging population.

Some critics have said that the benefits of human genome sequencing and functional genomics have not been fully realized yet. We need to work beyond traditional diagnostic tools, and this is exactly the challenge that lies ahead for Affymetrix and other companies. We need to work collaboratively with academic and clinical researchers, physicians, and drug makers to create powerful tools to query the genome, identify informative biomarkers, create precise, cost-effective molecular tests and companion therapies, and ultimately, improve patient outcomes. The tests need to be sufficiently simple so that their use extends beyond the handful of major medical research centers to local healthcare clinics. The field is still in its infancy, but the potential to impact the quality of life, support reduction in healthcare costs, and enhance our understanding of disease is enormous.

Ultimately, the Genome Generation is about you. I would welcome hearing from you and understanding your story in the Genome Generation.

¹ Lander E. S. et al. Initial sequencing and analysis of the human genome. Nature 409(6822):860-921 (2001).

²Venter J. C. et al. The Sequence sequence of the human genome. Science. 291(5507):1304-51 (2001).

³The Genome Generation, a book by science writer and molecular biologist Elizabeth Finkel, tells the stories of scientists all over the world who are part of the genomic revolution.