International conference on quantitative genetics 4: big science for complex traits


Five years after the very successful and inspiring meeting in Hangzhou, China, the quantitative genetics community reconvened for the 4th International Conference on Quantitative Genetics to a meeting entitled ‘Understanding Variation in Complex Traits’. On 18 June 2012 at 8:57 am, 3 min ahead of schedule, ICQG4 ( was kicked off by Bill Hill in one of the cradles of the discipline: Edinburgh. The conference lasted for five full, exciting and exhausting days and was attended by 670 delegates from 46 countries. It was run as a series of plenary sessions lined by hundreds of posters. The whole Wednesday afternoon was reserved for presentations of postgraduates, followed by a ceilidh (the traditional Gaelic social gathering party). Throughout the conference, there was time for chatting, networking and meeting with colleagues, a fundamental component of scientific exchange.

A whole list of influential consolidated speakers was punctuated by promising younger researchers in an aggregate that would need time for digestion. Quantitative genetics is the analysis of complex traits, and this vagueness permeated the whole conference, with the pleiotropic effect of a highly diverse and diversified meeting. Despite this, some bordering disciplines were (almost) absent: Bioinformatics, Population and Conservation Genetics, perhaps due to an overlap with the neighbouring Molecular Biology and Evolution meeting, which was held in Dublin the following week (

During the conference, the Mendel Medal of the Genetics Society was awarded to Eric Lander. He presented an excellent review of the exciting and seemingly ever accelerating development of technologies, concepts and achievements over the last decade following the landmark publication of the draft human genome sequence in 2001. If only for listening to him, the conference was worth the trip.

There is a general agreement that quantitative geneticists presently live in exciting times: huge genome-wide association studies (GWAS) in humans, large-scale genomic selection schemes in farm animals and crops, as well as in established model species like Arabidopsis, Drosophila, C. elegans and mouse (just to name a few), high-throughput genotyping, expression profiling, sequencing and parallel computing. However, Daniel Pomp rightly reminded all of us that the community was in the exact same mood of enthusiasm some 20 years ago when the first microsatellite-based mapping experiments were conducted and many predicted that the full understanding of the genetic basis of complex traits was just around the corner.

Have we come any closer to this ultimate goal? in some sense, we have, since many of the open questions people worked on twenty years ago today can be satisfactorily answered. But do we really understand any better how complex genetic variation is built, how it works, and how it has evolved? Or are we in an ‘Achilles and the tortoise’ set-up where the target (the understanding of the genetic basis of complex traits) moves a step away whenever we come a step closer? We have certainly learned with some humility that the genetics underlying complex traits is far more complex than we ever thought it would be. A complexity that endlessly unravels as new technologies, such as epigenomics, emerge.

Another general impression is that data has become extremely large and complex, yet the development of statistics does not always appear to keep pace with this development. For instance, a brilliant talk was dedicated in its entirety to principal components, invented by Karl Pearson back in 1901 and still fully operational nowadays. A significant proportion of talks, though, for example many of the GWAS presented, use statistics that are not much beyond basic regression techniques. Overall, few slides in the talks were dedicated to methods themselves, and the purely methodological talks were a minority. This is not because complex statistics is not being developed. The reason, or so it seems, is that producing data has outpaced expectations by many orders of magnitude. Take next-generation sequencing data as an example. It is hardly possible to continue storing and analysing data at the pace they are being produced, even in the cloud. The widely expected advantages of large-scale sequence data in terms of, for example, improving accuracy of genomic selection or power of association studies, in general are not confirmed by the preliminary studies presented at the conference.

All this has resulted in quantitative genetics becoming Big Science, where huge resources are allocated to extremely expensive experiments and studies, especially so in human genetics. The scene is dominated by large-scale exploratory studies (most prominently, GWAS based on tens of thousands of individuals genotyped for hundreds of thousands of SNPs) while well-designed, hypothesis-driven experiments aiming at answering pertinent questions are rare. An exception was the thoughtful, though provocative talk by Trudy Mackay presenting substantial empirical evidence, which suggested a prominent role of epistasis in the inheritance of complex traits in Drosophila. This (like some other talks) triggered some controversy in the plenum as well as in the aisles.

The conference was sparse in strict animal breeding topics. One of the best days for animal breeders was Friday, comprising sections dedicated to interaction between individuals and to genomic selection. In the former session, Bruce Walsh brilliantly replaced Piter Bijma, and the session moved from theory to practical applications seamlessly. In the latter, a much anticipated talk by Ben Hayes reported the first, preliminary analyses of the impact of having full bovine sequence data on improving the accuracy of genomic breeding value estimation: so far, modest.

During the meeting, we learned about the sad and premature decease of George Casella, eminent statistician at the University of Florida who had made decisive contributions to widely used techniques in our field, such as introducing the Bayesian Lasso.

It was agreed that the next ICQG shall be held in Madison, Wisconsin (USA) in 2016.