Well, the American Society of Human Genetics Annual Meeting is coming to a close for another year. My talk is done and dusted, so I no longer have to lie awake at night worrying that I will forget everything other then the words to “Stand By Your Man” when confronted by the crowd. My white suit is now more of an off-white suit, with regions of very-off-white and pretty-much-entirely-out-of-sight-of-white. I’m looking forward to getting back home to catch up on my sleep.
For the last time, I’m going to give a little summary of talks today that I thought were interesting, or gave some indication of where genetics may be heading in the future. I will write up some more general thoughts about the meeting in the next few days, as soon as the traveling is out of the way and my mind has recharged.
If you would like some second opinions on the conference, GenomeWeb has a number of articles, including a couple of short summaries, as well as a nice mid-length article about the 1000 Genomes session; there are also a number of articles over at In The Field, the Nature network conference blog.
Measuring and Quantifying Relatedness
There was a lot of material in the first session of the day, and it was pretty diverse in its scope. A number of people (myself included) talked about impuation, and how that can add value to old data. But I won’t talk about that, because after live tweeting my own talk (1234) if I massage my own ego any more I will probably go blind.
One thing that was cool was the reporting of a number of methods for identifying and surveying Identity By Descent (IBD). Bruce Weir gave a cool talk on theoretical considerations in IBD, and gave some maximum likelihood estimators for IBD parameters, as well as their variances; these give a way to map IBD across the genome, and to estimate the relatedness between individuals in a precise way (to distinguish between, say, brothers, half-brothers, parents and cousins).
Sharon Browning talked about extending the phasing algorithm used in the imputation program BEAGLE to simultaneously do phasing and identifying IBD; by the looks of things, the method gives very impressive resolution, and seems to capture virtually all IBD greater than 3 cM. Given Brian Browning‘s talk yesterday on extending BEAGLE to do genotype calling, it appears that soon all of bioinformatics will occur by typing
java -jar beagle.jar –do SCIENCE
Active Subjects in Genome Research
A bit of a theme in a number of sessions I saw today was the idea of moving away from the old passive model of research subjects (data is collected from them), to a new active model (they know about the research, volunteer information, decide who to share their data with).
Holly Tabor talked about an in depth study of exactly what the expectations, hopes and fears were for families suffering from a Mendelian disorder, as they entered a whole-genome sequencing study to track down the cause. The talk was illustrated with lots of quotes from the family; what was surprising was how excited they were to be able to understand their disease and help other families, and how the importance of the dreaded privacy issues paled in comparison.
Ingrid Holm talked about the Gene Partnership Project (GPP); an exciting initiative to combine children’s medical care and phenotype research. Participants are patients at a children’s hospital, and are all genotyped at birth; their genotype data and medical data are then kept in a joint clinical/research database. What I found cool about this talk was the clever system they put in place to keep patients involved, while maintaining anonymity and ensuring that patients can decide what information they do and do not want to receive.
Finally, the masters of Active Participant research, 23andMe (embodied, in this instance, by Nick Eriksson) weighed in, presenting a load of gene associations for odd phenotypes (hair curl, eye colour, that kind of thing), based on DTC companies’ customers voluntarily filling out web forms. 3 of the studies produced new, significant hits, and many existing hits were verified. I was impressed by the lengths that 23AndMe went to to ensure a lack of bias in their questionnaires, including asking the same people the same question in different ways at different times.
I am far from an ethical or legal expert, but it seems to me that by engaging your participants and giving them a stake and a place in the research project, you get around a lot of the classical problems of consent. Subjects that are in constant contact, learning from the research you produce and wanting to get involved more, can always be asked for more information or permission to study new things, in the knowledge that you will get a good turnout. This allows studies to roam around, add new aspects or get new value from old data, without abusing the trust of the people you collected the data from.