Author Archive for Konrad Karczewski


Personality and Genetics

These days, it’s rare to come across a published study that describes a negative result. It’s even rarer when lots of people are interested. But keeping in the theme of missing heritability, a report in Biological Psychology was released a few days ago that found no common genetic variants significantly contributing to differences in personality. This should not be surprising for anyone who has been following the ongoing search for heritability. If height (a definitively heritable trait with an easy-to-measure quantitative output) cannot be easily explained by common variants in large studies, a search for genetic associations with less objective trait such as personality is certainly in trouble.

And that’s exactly what happened. “No SNPs reached genome wide significance (α = 7.2*10-8) and the SNP with the lowest p-value for each personality scale explains less than 0.5% of the total variance.” The usability of the psychological tests (and any personality test, for that matter) for a quantitative measure of personality is a contentious topic, discussed here and here by psychologists in the study’s wake. Boiling down an individual’s personality into a series of numbers is not a perfect science (“perfect science” is sort of an oxymoron in itself, isn’t it?) and one that would no doubt change over time (I had friends that took Myers-Briggs tests yearly to look at how the results changed). In any case, even if this sort of information could be easily ascertained, the discovery power will likely be low at current GWAS power levels. Considering personality as a “rare disease” (in one sense, every individual has a basically distinct personality, so it’s the rarest trait around), finding the (likely rare) variants that contribute to personality will indeed become a challenge.

It’s here where larger aggregated studies might come into play. Citizen science groups like DIYgenomics and research snippets/surveys in 23andWe can theoretically open the door to greater amounts of data than any single researcher can afford. Of course, such endeavors are subject to some limitations. Most importantly, traits in these citizen studies are self-reported. This means that self-perception will alter the results (whether consciously or sub-consciously). For instance, if a 23andWe research snippet asked “Do you consider yourself a good person?”, citizens’ perception of themselves will most likely steer the answer in a positive direction (although it may be an interesting experiment in genetic associations with self-perception itself, in observing the individuals who answer “No”). Personality test questions are often more advanced than this one, but may suffer the same problems.

Thanks to Razib Khan for getting this discussion going. It will be interesting to see where the investigation into the genetics of personality goes next.


Response to GAO testimony on DTC Genetic Testing

This was a guest blog post I wrote for Russ Altman over at his blog:

On July 22, 2010, the Government Accountability Office (GAO) released a testimony on Direct To Consumer (DTC) Genetic Testing companies. In the testimony, the GAO sent DNA from 5 anonymous donors to 4 anonymous DTC companies. At times, the results were astonishing, such as claims made by some of the companies taking advantage of ill-informed customers to sell custom supplements “based” on genetic test results. However, the testimony also revealed a fundamental disconnect in communication between science, medicine, and the public: a disconnect that has always existed, but is now being brought to the public eye, as recent technologies have begun to bridge the gap between scientists and consumers.

To preface, it is of course outrageous that anyone interpret a DTC genetic test as a diagnostic test (at least in their current form). Analysis of a personal genome is not a medical test. For the bulk of genetic markers, having a “high risk” allele for a disease is not even close to a diagnosis of the disease. It is simply an indicator that on average, in the particular population chosen by a research study (which are often small populations or populations selected to be enriched for a particular disease), individuals with that particular allele had a higher incidence of the disease in question than individuals without the allele (i.e. the “high risk” allele has a higher odds for the disease than the other). The companies then translate into a overall disease risk, which adjusts the prior probability of getting the disease by this odds ratio. Depending on which studies and genes/alleles a company takes into consideration, this risk may be vary considerably. In any case, the report provides a final probability of getting a disease, which may or may not actually be the same as the actual outcome. Just as an individual can get lung cancer without smoking, one can get diabetes even with a below average risk.

This is not to say all the calculations of the disease risk interpretations of all these companies are flawless (we haven’t verified the math and studies in all these companies), but the fact remains that there are legitimate scientific differences on how to interpret the data. While no particular method is outright “wrong,” there are better and worse ways to analyze results of genetic tests and competition among DTC companies for the highest quality interpretations should become increasingly important. Of course, it is objectively difficult to measure which interpretation is “best,” but this will change as more data become available both in predictive claims and possibilities for validation.

According to the testimony, the Department of Health and Human Services’ Secretary’s Advisory Committee on Genetics, Health, and Society notes that “[practitioners] cannot keep up with the pace of genetic tests and are not adequately prepared to use test information to treat patients appropriately.” While this may be true at present, this need not stop information from genetic tests from entering the clinic. A general practitioner may not be able to keep up with the latest advances in neurosurgery, but that’s where the specialist system thrives. In any case, just as clinicians are expected to demonstrate a basic level of competence in immunology in medical school, genetics must be treated the same way. Here at Stanford, a pilot project was launched to teach medical students about the field of genetic testing in an interactive classroom setting with state-of-the-art methods for analysis of personal genotypes.

Deceptive marketing, including “personalized supplements” (allegedly with celebrity endorsements) and drugs that may “repair damaged DNA” (allegedly called “epigenetics”), to say nothing of surreptitious testing and scientifically nonsensical claims, are inexcusable and irresponsible practices for any company, not limited to this particular market. However, the delicate matter of genetic testing and its use as a clinical guidance tool is a concept that must be explored further. The GAO uses the phrase “misleading test results”: it should be noted that while the current implementation of the reporting of test results may be in certain ways misleading, the framework of genetic testing is not in itself misleading. Proper interpretation is based in the same mathematical and biological context as much of today’s medicine. There is great potential for the use of genetic tests in the clinic, so long as results are carefully interpreted. While this was often limited to geneticists in the past, we hope that this can be soon accomplished by physicians and the public.


I’m here

I figured it was about that time that I start a blog and a general information page so unsuspecting tourists on the internet (whether they were looking for me or not) could find me. Well, as promised, here I am. You can find out more about me, my research, or find collaborators/old co-workers and things I think are cool.

I’ll post some interesting perspectives and news on this blog once in a while, but I can’t promise any regularity. For now, enjoy the ride.

RSS Recent Publications



Get every new post delivered to your Inbox.