genomics & society

Pace of Human Sequencing Far Exceeds Everything Downstream Including Genetic Counseling

Posted by Myra I. Roche on December 2, 2011

While keeping in mind the dictum that correlation does not imply causation, yesterday’s New York Times reported two trends that are pretty clearly causally related: the number of human genomes expected to be sequenced by the end of the year (30,000) and the precipitous plunge in the cost per genome ($8.9 million in 2007 to today’s bargain basement price of $10,500). According to the article, China’s BGI,  the world’s largest genomics research institute, is churning out 2,000 human genomes a day. The world’s capacity to generate sequence was quoted as 13 quadrillion (13,000,000,000,000,000) DNA bases per year.

Data Delivery by Rickshaw?
But while sequencing has quickly transitioned to a high-throughput operation, downstream events, like storage, interpretation, and communication of results, have been unable to ramp-up enough to cope with, let alone keep up with, this deluge of data.  The Times article focused primarily on storage capacity and retrieval painting a quaint image by revealing that BGI transmits the data back to their clients using, not the poky Internet, but rather, on discs, using old-reliable, FedEx; presumably brown shorts and all (or is that UPS?)but, hopefully, not via rickshaw.  Remember when Craig Venter and James Watson’s genomes were first reported in 2007?   Watson received his genome, although not including his APOE status which he chose to have redacted, on the 2007, state-of-the-art storage unit, a miniature hard drive.  So much for the transition to high-throughput.

Genetic Counseling of Sequencing Results
Those in-the-know express confidence that the storage problem will be solved by a combination of outsourcing the problem to, say, Google, and by storing a lot less data per genome than is currently done. But interpreting the data, and combining it with emerging information from RNA, proteins, and other biological data, has created a huge bottleneck. The biggest bottleneck, however, may be in the communication of results.  In their 2010 article, Ashley and colleagues reported the amount of time it took to interpret the sequence from a single individual with a family history of coronary artery disease. They focused on four areas of interest including variants associated with known Mendelian disorders,  novel mutations, variants associated with pharmacotherapy, and SNPs associated with common diseases.  In the accompanying viewpoint paper, Ormond, et. al. estimated that it would take at least 5 hours of direct patient contact to communicate the results.  This marathon counseling session did not include the multiple hours spent preparing for it.  Obviously this is not a viable option for lots of reasons including cost, the limited number of trained professionals, and, if nothing else, the limited attention span of the average person.

Counseling Dr. Watson
When Watson’s genome was published, I wrote an opinion piece imagining what it would be like to counsel him if he just happened to stumble into our clinic by mistake with miniature hard drive in hand and compared the process to counseling someone that I defined as perhaps more representative of the general population.   While in some ways, counseling Dr. Watson would be easy (for example, probably don’t need to worry too much about obtaining truly informed consent) still, there would be problems.  For example, his sequencing (admittedly based upon the NCBI build 36 not the current build 37.3) predicted he would have both Usher syndrome type 1b and Cockayne syndrome.  Wrong.  But at least these changes produce recognizable phenotypes and, while he clearly has neither, what about erroneous predictions that are not so easily dismissed?  For example, the report mentioned in my previous post about the NIH Undiagnosed Disease Program quoted an error rate of sequencing of 1 out of 500 bases.  One patient generated over 380 gigabases of sequence data with roughly a 126-fold (theoretical) redundancy across the genome.  While “deep sequencing”, sequencing the same region multiple times, can address this problem to some extent, errors are not uncommon partially because of gaps in coverage.  Perhaps a better approach is to report the fraction of the genome/exome that is “callable”; that is, when all filters are applied, “what fraction of the genotypes can be determined at a specific confidence threshold“?  An article by Ajay, et. al explains this concept by comparing the same clinical sample sequenced on two related Illumina platforms with 126× coverage and can be found here.

Is Genetic Counseling Scalable?
Still, even assuming interpretation methods will become more accurate and more streamlined,  genetic counseling is likely to stubbornly remain the last true bottleneck seemingly impermeable to a high throughput process.  Extrapolating from the 5+ hours/per sequence estimate, it seems clear that the current model of genetic counseling must change, but how?  In the past, genetic counseling has evolved much like a population of organisms; by simply piling on new functions (new genes, new tests, new risks) as they arise without bothering to jettison any of the antiquated components.  But the true weight is finally catching up with us and threatens to swamp the entire ship.  It remains to be empirically determined which parts of the genetic counseling process are comparable to the exomes, the vital coding sequences of the communication process that must be preserved and maintained, and, which parts are more akin to the introns, parts that must be excised and thrown overboard.  Or could some alternative splicing be in order?  Alternatively, perhaps new models of communications can be allowed to emerge that may function better, unencumbered by the weight of the past.  We might even have to change the name to “genomics counseling”.  At any rate, the conversation to explore these possibilities is long overdue because running as fast as we can just to stay in the same place is no longer an option.


4 Responses to “Pace of Human Sequencing Far Exceeds Everything Downstream Including Genetic Counseling”

  1. Wendell said

    Great post, Myra.

    “What would this new method of communication look like”? Initially, I thought “Genomic Interpretation Centers” or “Genomic Hotlines” could be created to assist in conveying information with regard to one’s DNA profile. But, afterwards I began to think about the ELSI issues that may arise. Like other ELSI issues I don’t think there is an easy answer to the above stated question. However, do you think it would be wise (or even practical) to effectively train individuals in the area of “genomic interpretation” at the undergraduate level to help (i.e. function as genetic counselors) address this issue?

    • Myra I. Roche said

      Thanks for the comments by Wendell, Martha and Karey. I agree that the traditional face-to-face counseling model is not sustainable. There are existing models of phone consultations with certified genetic counselors (both pre-test and post-test) and this has become more accepted but, in general, these are not reimbursable services thus limiting their current use. While the scarcity of genetic counselors has received a lot of attention, one big problem is their uneven distribution. For example, at last count, there were 35 genetic counselors employed by UNC-CH and Duke alone. Many of these counselors are in research positions and are not providing clinical services. Of course, geneticists are even less broadly distributed. Computer delivery of information has been tried and shown some success. Its primary advantage is that by taking over the background information building, it frees up the genetic counselor to do what people do best; the personalized delivery of information and counseling. Since much of genomic information has health implications, the requirement of certification (and, in some states, licensure) serves to protect the public from harm. Clearly a triage system of sorts is needed as not all genomic information has the same implications and just because something is genetic doesn’t mean a genetic counseling consultation is automatically warranted.

  2. Martha K. said

    In considering “this new method of communication,” I wonder about the ability to harness technology as a tool for making the workload lighter. Would there be some segments of the counseling that could be done using computer modules? They could even provide feedback or a kind of quiz at the end to evaluate how much the patient understood and help reinforce the messages. I certainly don’t think the counselor should be removed from genetic (genomic?) counseling in general, but perhaps a computer counseling session could be applicable in some cases? Just like with Wendell’s thoughts of a hotline or center, this would create its own set of ELSI issues– particularly in the realm of data security and technological literacy.

  3. Karey said

    I agree with the above comments. The identity of the genetic counselor is changing with with the rapid advancement genomics research. I wonder how current genetic counselors feel about this change? Yes there is a call for new modes of communication, but I am not sure exactly how that would look. I do agree with Wendell, that the training program for genetic counseling would probably have to change to incorporate these new advancements. Additionally, with the small number of current genetic counselors and their increasing workload, may push medical consumers to look elsewhere for answers and interpretation of genetic information which could be problematic.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s