Two rarities seldom seen by Bostonians are the American Society of Human Genetics Annual Meeting (aka ASHG) and the baseball World Series. The former was last in Boston 60 years ago, in 1953 – the year of the double helix. The latter, a contest between grown men – as evidenced by a dazzling roster of beards – playing some sort of rounders derivative, has not been won on home turf by the city's Red Sox since 1918. But both events converged this year, with the geneticists of ASHG more than equal to the task of keeping all four bases covered. That is, the DNA bases A, C, G and T.
Mo' data, mo' solutions?
The most inventive session title of ASHG 2013 surely goes to 'Mo' data, mo' problems?' But a bigger theme of the meeting was how we are only now beginning to hit the sample sizes needed for addressing many of the most prescient questions in human genetics. Newly available datasets boasting tens of thousands of exomes have given much needed power to the analysis of complex trait heritability, and broader sampling of global populations – let's call it the 'humanome', why not! – together with n > 10k has generated the landscape of rare variants necessary for deconstructing demographic histories.
The extent to which these variants, being as they are collectively very common but in each instance vanishingly rare, contribute to complex traits is an important question, and one that it is hoped the data expansion will eventually solve.
Mo' than a feeling
What some may view as an embarrassment of riches when it comes to data for others is just… well, an embarrassment. Relying on less than robust stats, many 'causal' variants had been claimed from GWAS and other omics studies that even their very discoverers are now skeptical about: intuition and a trend, it seems, are not enough. In a session focused on the logging of clinical variants, one questioner even asked what facilities databases are going to provide for repentant researchers wishing to retract claims of disease association for a given locus.
Even where intuition does not make mistakes, it can narrow our scope. Marc Vidal showed that protein-protein interactions reported in the literature, which can be biased for binding events tested by researchers who felt they were likely to occur, do not exhibit any special enrichment in his systematic proteome-wide survey of interactions.
Mo' functions, mo' phenotypes
Robust statistics for an association is the first step, but genetics really comes alive when we can point to the functional consequences of a variant. And more and more groups are seeking to do this, both through the examination of functional genomics and through phenotyping model organism knockouts.
Where ENCODE data had taken center stage at previous ASHG meetings, a new kid was on the block this time around: the GTex consortium, whose data matches genotype to gene expression and in doing so catalogs eQTLs at a vast scale. When complemented by other regulatory data, such as that produced by ENCODE or the Roadmap Epigenomics project, the impact of variation on gene expression – as a first step to function – can be quantified.
Model species enable the easy manipulation of genes for whole-organism phenotyping in a way that would not be deemed ethical in human populations (ASHG's session on gene therapy notwithstanding), and in recent months their study has benefitted from powerful new genome editing tools derived from the CRISPR/Cas9 system.
However, human geneticists dissatisfied with making cross-species assumptions, and too impatient for the large sample sizes needed for convincing GWAS, have a get out – a bunt, a pinch-hitter: twins. In a session bookended by Walter Nance and Tim Spector, the conference heard how twin studies allow human traits to be studied with drastically reduced sample sizes, to the extent that new data generation has focused on increasing the number of data types rather than the number of source individuals.
Even mo' data
The ascendancy of single-cell studies is a notable trend within genomics. Aviv Regev and Garry Nolan closed the meeting with presentations demonstrating the importance of single-cell approaches to truly understanding the molecular biology of cells, with an instructive example of bimodal expression patterns that are averaged out by traditional methodologies. Of course, single cells present us with a data problem, as not only are we now saying that tens of thousands of individuals must be sampled, but also large numbers of cells from within these individuals.
But breaking down an individual to multiple genomes is crucial for understanding human disease, the conference heard, because somatic mosaicism is rampant in cancer and also common in complex genetic disease, as shown respectively by Nazneen Rahman and Debbie Nickerson.
Is there a gene for belief in genetic determinism?
The St Louis Cardinals, the Red Sox's World Series adversaries, were not the only Midwesterners in town for ASHG, thanks to a pitstop from controversial comedienne Kathy Griffin. Griffin once asked presidential contender Michelle Bachmann whether she was 'born a bigot' or instead 'became one', and so clearly has a longstanding interest in the heritability of complex traits.
But, if there were a gene for bigotry – and anti-genetic determinists would howl at the very idea – should we test for it, or even seek to cure it? Ethical questions such as these, although I will allow for the most part less hypothetical, always form the heart of ASHG, which is notable for a significant clinical attendance.
Geneticists at the coal face (that is, who use genetics to treat real people) tend to view the subject through a prism dominated by ethical quandaries. But that is not to say that researchers outside of the clinical sphere cannot also be inspired by ethics.
Step up Nathan Pearson, who is developing a collaborative personal genomics platform. Pearson mercilessly paraphrased Boston native JFK to exclaim: 'ask not what your genome can do for you, but what it can do for society!' There wasn't a dry eye in the house.