Personal sequencing

What's your comment on personal sequencing, i.e. sequencing your genes to see the risk of getting certain diseases, something 23andMe and deCode are doing?

I've been thinking about doing this pretty much since the first service became available (from DeCode Genetics - DeCodeMe). The main issue I always had was the price (over $1,000 just a year ago). Then 23andMe dropped their price to $399 last year which earned them the Top Innovation of the year by Time magazine. Since then everybody else has started to drop their price, including Navigenics which started at almost $2,500 and now is under 1,000.

So if price is not an issue, go for it. If you are like me though, you must carefully consider what you are buying with your money. First of all, these services are not DNA sequencing but genotyping. They characterize a few thousand to just over a million SNPs from your genome using a microarray. Then they use that information and compare it to the less than 300 genome wide association studies performed so far that tried to link a disease to genetic signatures (SNPs). In other words, while you will get publication-quality SNP data (assuming the company you go with actually gives you the raw data, which is not true for all providers), the information available linking that genetic information to actual diseases is very limited at best.

Currently there is a study going on at the Scripps Institute, in collaboration with Navigenics, to determine just how pertinent this service really is. There are examples of this information being very useful clinically, and there are sample showing that this information is basically useless.

Personally, this is what I think: while DeCode was the first one to jump into this industry, they are not doing very well (not too long ago they fired a good chunk of their scientists just to keep afloat). 23andMe, while well funded (having the Google connection), has a website that leave a lot of questions unanswered (I am a Geneticist; if I do not see the basics of how you generate your genotyping data I cannot trust you). In other words, 23andMe is focused more on the commercial aspect, trying to make themselves "cool". And there is Navigenics, who right now have enough bioinformatic information to provide you with an estimate on the likelihood of developing about 24 diseases. If I were to spend a few $100 on this service right now, I would go with Navigenics. I would not spend >$1,000 in it though since the prices will continue to drop. 

FYI: Complete Genomics has made a claim that they will start selling a service this summer where they will sequence your complete genome for $5,000. That is hard to believe but if at least part of it is true, then you will see the prices of these services go down even more.

Thats funny. Unlike Ivan, I haven't given this a second thought until about 5 min ago.  I guess it sounded to "GATTACA" to me.       

So for me it comes down to how truly predictive these studies are for an individual, i.e., their track record.  If people with SNPs predicted for certain mutation causing disease, begin saving themselves 20 years of misery on a regular basis, it might be the best $1000 you ever spent.  That said, I think we are a few years away from that kind predictability at least as far as whole genome sequencing goes. The whole complete genomics sounds like a bit of a farce to me.  I seem to remember that the human genome isn't even truly complete?

Of course if I knew a relative carried deleterious mutation and I was a potential carrier, then Id probably have already done it.

What is your take Pharmtao?

Rus

With regards to the Complete Genomics, I believe the cost is currently $20,000 per genome (with a minimum of 5 genomes).  Not exactly affordable, nor is it likely to be desirable for the average person.  What they hand you is the entire genome (essentially hundreds of millions of aligned short reads).  It would be a daunting (not to mention computationally challenging) task just to make a comparable interpretation from this data.  In addition, if one is interested in disease associated SNPs, then he/she would likely end up resorting to the known SNPs (i.e. what is called by a genotyping array) to determine the likelihood of developing a given disease.  Perhaps once enough genomes have been sequenced with short reads (i.e. the 1000 genomes project and beyond), we may be more ready to handle such datasets in a more personal setting.

The idea is very catchy for people that know little about how it works. One can show rosy scientific facts and prove it works  well.

Does anyone knows the success rate of this commercial venture? I mean how many predications came true?

Its too early to say since there has been no systematic study of the results (yet at least one such study is ongoing). At a recent conference (Scripps) an example was given of an individual with perfect health (and younger than 40 years) who nonetheless received a prognosis of a high probability of developing prostate cancer. As a result of this he went ahead and got a PSA test and got a very high score, which eventually lead to chemotherapy.

Such an example is as much marketing as any else. The bottom line as pointed out already it is very early for this technology, and like many cutting edge application it all "depends". In other words, it literally could be useful for you or it could be close to a complete waste. It is somewhat like when the first iPod came out (if you recall). It was nothing like the ones today and the music and applications were very limited (it actually had a game, but at the level of Pong). Today you have the iPhone, the App store, etc. These tests have a use, but very limited. The longer they are around the more useful they will became. 

One could make the point that knowing the order of the letters doesn't help unless you speak the language, ie where to words and sentences begin and end.  We still have a lot to learn about this level and so maybe it would be better to wait until at least the experts understand the lingo.

Then there is the problem of epigenitics - much of the expression of our genes is not simply controlled by the sequence of that gene, but by a host of other mechanisms that we are just beginning to scrape the surface of (so again- one day probably, but not today). In addition to epigenitics (methylation, chromatin modifications etc) there are also the microRNAs that we've only just discovered - imagine that !

Then, you ask yourself, who is doing this sequencing and how safe is the data? Is the sequence data treated as  a patient's medical record - if so then I'd suggest that private companies should have no right to hold that data as they are not bound by a hypocratic oath that prevents them doing things with the data that is not in the patient's best interests and with the patient's consent.

Maybe finally, I think there should be a support system for patients / customers of such a service that enables them (the customers) to a) understand the data (which is difficult since the professionals are still trying to figure it out), and b) advise them about potential routes to mitigate potential disease risks that are flagged up - if this is even possible. Having a collection of gene mutations that are common in people who go on to have high blood pressure might be useful and lead the customer to change their diet, but if there is nothing the person can do then a would suggest that he/she would be more likely to suffer from anxiety and depression as a result. This would clearly be a poor outcome if the likelihood of the disease actually aflicting that person was less that 100%. Either way, medical support would be required and I'm guessing that private sequencing companies are not providing this in their $399 service.

We also have to consider the insurance ramifications from personal genomic data. If the info is available, a potential insurance company could deny you coverage because of your genetic predisposition to some sort of expensive disease. The treatment of personal genetic info is a hot topic in the field of bioethics. Here is a link to another discussion on the topic.

 www.scientistsolutions.com/t7235-personal+genomes+available+online.html

www.scientistsolutions.com/t5669-dna+testing+and+insurance.html

Omai

 I just added this topic to Digg.  (http://digg.com/health/Scientist_Solutions_Personal_sequencing)

Seems like it might be pretty important to those considering these types of services to hear from us scientists about it.  I encourage you all to spread the word a bit. 

Rus

I agree that just getting your genome sequenced at this point is close to useless (you would have to do a ton of bioinformatics works to figure out what each SNP means - close to impossible for even molecular biologists). Much of the value of these genotyping services (most of the commercial services like 23andMe and Navigenics are actually SNP genotyping microarray experiments, not DNA sequencing) is the bioinformatics they provide (algorithms linking the results of medical studies to particular SNPs). DeCode has one of the best databases due to their years of work cataloging the Icelandic population, followed by Navigenics and 23andMe. Yet even this information is limited. I would venture to estimate that the results you get are akin to running a Western and funding that there is more of a protein in your untreated versus treated controls. Promising results, but just the first step in a series of experiments to determine if this observation is really true. These tests provide a glimpse of an array of genetic conditions. Anything you obtain from them could offer you an opportunity to investigate further.

I also agree that for most people this service would likely be of little use, not just because of the extra expense and the high likelihood that nothing will show up, but also because of the limited scope of the diseases screened (my very rough estimate is that only 5 to 10% of what can afflict you is being screened for - and I am being optimistic). Yet, if you are curious about your genetic makeup and like Google says: if you are "Feeling lucky", this may give you some useful information. These services are at a stage that can be clearly defined as "early adoption".

As a side, some of these services do provide additional support. For example when you get the Navigenics genotyping service, it includes a one hour discussion with a certified genetic counselor who will explain to you what your results mean. Also you are given the option to remain subscribed to their service (at a $100 or so fee per year!) and receive updates whenever new connections between SNPs and diseases are made. Finally, all this information remains 100% confidential so there is no chance insurance companies would gain access to it.

About those 300 SNPs, how reliable their results are?  As I know, only about 6% of association studies have been replicated.  Have the results of those 300 SNPs been replicated?

It is not 300 SNPs, but 300 GWAs (genome wide association studies). By definition a GWA study compares a given disease with healthy controls using some kind of whole-genome technology (typically a microarray containing a million or so SNPs). As for reliability, that is a big question. Just because a GWA study is performed does not mean that relevant association data was generated. In other words, no SNP may have been identified that was statistically relevant to the disease. This is why most of these studies end up associating SNPs to a given percentage of the disease (for example, SNP #453 is associated with 3% of disease X). As you pointed out very few of these studies have been replicated, to a great extent because of the large expense and also because many other novel studies still need to be performed. 

It seems ok  for one to be really curious to know the genetic predispositions to certain diseases so that one can make changes in his/her lifestyle so as to avoid the disease. But personally, I would be shit scared to do so. What if i find that I have a predisposition for something inevitable as Hungtington's. Rather than spending my life normally, i would be just traumatized by the very idea.

I am totally in for the genetic testing of embryos coz then its the parents who have to make the decision, but testing after one is born just seems like a burden that one is forced to bear when they dont really have to ....

Knowing that you have an expanded Huntington allele would be really devastating, however, it i useful to know if you have risk alleles for other diseases.  My favorite example for this is Hereditary Hemochromatosis.  If you know that you are homozygous for two of the common disease alleles, you can prevent the course of the disease by donating blood to alleviate the build-up of iron in your organs.  In this scenario, knowing is better than not. 

I don't think there is a universal answer for this topic... it must be considered on an allele by allele basis.

Thought you guys might be interested in this article that appeared in the NY Times about the topic of personal sequencing.

www.nytimes.com/2009/04/16/health/research/16gene.html

It seems the big gun geneticists are not sold on the idea either!

Rus