How much fun is a person allowed to have when going to a conference? That kind of sums up my personal experience going to the first ever Capita Selecta in Complex Disease Analysis (bi-annual) conference in Leuven, Belgium. Not only was it fun, but the conference program was packed with interesting talks and short courses. Dr. Kristel van Steen and her team did an excellent job organizing the meeting, with plenty of time for socializing and networking among the participants, including a guided tour through the city of Leuven in 3 courses (meaning: during the tour we visited 3 different restaurants). Did you know that the city of Leuven already had a university in 1425? This kind of history permeates Leuven, with its well preserved 17th century Grand Beguinage (UNESCO World Heritage Site) an absolute highlight. As an American, you kind of marvel when you walk inside this 7.5 acre historic quarter. It feels like walking straight back in time.
Scientifically, the conference also had plenty to offer. For those not quite familiar with GWAS, there were 2 (well attended) short courses offered on Wednesday. Dr. David Evans from the University of Bristol did a great job explaining the intricacies of Design and Analysis of GWAS with SNP data. If you think GWAS will soon die an ignominious death, don’t hold your breath. GWAS will be with us for some time to come. What may change is how GWAS is conducted, in conjunction with Next Generation Sequencing, and how it will improve the effectiveness of these studies. BTW: did you know that the word CHIAMO, as in CHIAMO algorithm used for calling genotypes from Affymetrix arrays, comes from the Italian words ‘I call’? Yeah, CHIAMO is not the name of the author or some fancy acronym.
Another short course was provided by our own Dr. Christophe Lambert, who provided insights into the Design and Analysis of GWAS using CNVs. CNV analysis is not for the faint hearted, but it turns out that if you pay attention to study design and proper randomization of your samples, you may actually find something. (His presentation with PowerPoint slides was recorded and is available on our website here.) This may sound like common sense, but even today, when I talk with scientists, I cringe when I hear that (as funding permits), they are sending in ‘one plate of samples at a time’ to their core lab for genotyping. After listening to Dr. Lambert’s presentation you will definitely know what not to do, in order to avoid problems with downstream analysis of your CNV data.
Another hot topic during the conference was gene-gene interactions. This subject always seems to generate a certain amount of skepticism, because we already have a hard enough time finding the single gene main genetic effects in GWAS, let alone having to look at all pair wise combinations of SNPs on such a grand scale. But some interesting (although compute intensive) methods were discussed, including Multifactor Dimensionality Reduction (MDR) by Dr. Jason Moore from Dartmouth Medical School. Dr. Moore’s argument is that we should be seriously looking at these gene-gene interactions, because we are finding little as-is in these GWAS datasets. How practical it will be to apply these methods to GWAS or NGS, remains to be seen. As one of my colleagues opined, there is certainly gold dust to be found in examining the tailings of GWAS, but the main value is in finding the gold nuggets, which are the single gene effects.
What also caught my attention was that in quite a few talks Forests of Random Trees or dendrogram-like tree analysis was discussed as a possible way for finding SNPs associated with diseases. The company I work for, Golden Helix, has been using this type of recursive partitioning tree building since before I started working here, about 9 years ago. We find it especially helpful for building accurate predictive (disease) models and looking at gene-gene and gene-environment interactions and correlations. If you are interested in this type of research, let’s talk. Our multiple tree building package is flexible and highly accurate. It was kind of interesting to see that scientists at this conference (Vincent Botta, Olga Savenije and Victor Urrea) agreed on the usefulness of tree analysis as a tool for finding the genetic components to diseases.
Oh, and before I forget to mention it, the day before the conference started, Golden Helix organized a training event for our SNP & Variation Suite (SVS) software users at the University of Liege. During this training several non-SVS users showed up to hear what we had to say, especially on the topic of BEAGLE and BEAGLECALL. This 30 minute presentation by Dr. Christophe Lambert was recorded and is available, together with PowerPoint slides here. We find BEAGLE and BEAGLECALL extremely useful software tools for improving the quality of results for the association studies we conduct with our collaborators.
If you have wondered (like I have) how all these investments in genetic technologies are improving diagnosis and treatment of patients, the last day of this conference was a treat. Several presentations stand out, namely by Dr. Peter van der Spek from Erasmus Medical Center who discussed how many of these ‘omics’ technologies are helping with the diagnosis and intervention of patients with brain diseases (Development of the Face and the Brain: Towards the Understanding of Molecular and Cellular Mechanisms, Linking Genotype to Phenotype). Also noteworthy was a presentation by Dr. Severine Vermeire from University Hospital Leuven, who provided a case study on the genetics of Inflammatory Bowel Disease. Her hope is that a lot of this know-how will be used in a clinical setting.
All-in-all this conference was well organized and planned. Kudos to the organizing committee. I should report a potential conflict of interest, and that is that Golden Helix (the company I work for) was a main sponsor of this event. Nevertheless, it was money well spent. Thanks to Josh Forsythe, Jessica Vionas, and everyone else who helped organize this event from the Golden Helix side. … And that’s my two SNPs