ICHG2011: The state of play in complex trait sequencing

The International Congress of Human Genetics is rapidly approaching its conclusion (namely, my talk today at 3.15pm in Room 517, be there or be, I dunno, hanging out in Montreal I guess).

For me, the hightlight of yesterday was (somewhat obviously) the Individual Resequencing for Complex Trait Genetics session. This was organised by Mark Daly and Ben Neale of Mass. Gens. Analytic and Translational Genetics Unit, and gathered together a number of the Big Men (all men, unfortunately) of disease association together to talk about the many and varied Next-Generation sequencing studies they have been working on. I’ve summarised some of what was said below.

As always, you can find more coverage of ICHG on twitter (@lukejostins for me, #ICHG2011 for aggregated coverage).

The state of play in complex trait sequencing

Peter Donnely didn’t report on any of the seven common WTCCC diseases in his title (“Next-generation sequencing in seven complex diseases”), but he did talk about some pooled sequencing of DNA repair genes in 1000 breast cancer patients, to look for rare loss-of-function variants. He showed some of the difficulties involved in these sorts of studies, in particular the difficulties of calling small indels, and the radically different sets of indels called by different alignment and callng algorithms. Finally, he topped it all off with the sequencing of 500 clinical cases, and teling the story of how a young child with craiosynostosis (a skull full of holes) recieved a molecular diagnosis via exome sequencing, and how this diagnosis allowed his family to access support services that were denied prior to diagnosis.

Shaun Purcell reported on the sequencing of schizophrenia exomes, and demonstrated his amazing ability to take an apparantly null dataset and draw interesting biology from it (kicking and screamng if neccissary). He performed a case-control analysis, and found no hits, and a rare-variant burden test, and also found no hits. He then extended to testing individual pathways for burden, and found what looks liked pretty convincing evidence for an enrichment of rare variants in genes involved in synaptic pathways. The genes involved appeared to be an entirely different set to those involve in risk variants found by schizophrenia GWAS, by de-novo mutation mapping, and by CNV studies. Even cooler, he showed that the exome data can be used to give fine resolution to breakpoints in CNVs found by arrayCGH, and breakpoints that cut genes down the middle are enriched in cases compared to controls (despite showing no overall enrichment of CNVs overall).

Jeff Barrett, notre capitaine, presented on some of the various projects he was involved in. He started with discussing the immunochip projects: hundreds of thousands of variants in known or suspected GWAS hits for immune-related diseases, in hundreds of thousands of cases from dozens of consortia, in order to find new loci, replicated suspected ones, and fine-map well-established ones. He also reported on various family sequencing projects that our group is involved in, including as part of the UK10K project, and pointed people towards James Morris’ wonderfully geekily titled program Olórin for analysing such datasets. Finally, he hinted at a very large Crohn’s disease whole-genome sequencing project that is currently in the pipeline.

Chris Cotsapas reported on a big sequencing project of 1500 multiple sclerosis cases and 1500 controls, with the potential to follow variants up in the full 10K/10K of the International MS Genetic Consortium dataset. 35 candidate genes were sequenced, and so far about half the data is in. No solid results were in yet, but a few promising leads came up. Chris also noted that they also intended to do a pathway analysis, and then follow the results up by seeing whether mutation burden in implicated pathways had a effect on cellular phenotypes.

Finally, Gonçalo Abecasis (who always seems to be bought in as the “wild card” at these sort of conferences) talked about his projects using sequencing to examine quantative traits, and in particular LDL cholestorol. The most interesting project was the sequencing of an isolated Sardinian population, and using this as a dataset and imputation reference set to study quantative traits in the population. Somewhat magically, he found one of the fabled Low Frequency Variants of Large Effect, a 4% mutation in HBB, which in the homozygous form causes thallasemia, but in the heterozygous forms lowers LDL cholestoral by nearly one standard deviation. This mutation alone explains 8% of variant in LDL in the population. Even more impressive, this variant was entirely missed by GWAS in this population, as the chips totally failed to tag it. Even 1000 Genomes imputation failed to find it, only imputating using the Sardinian-specific reference set could uncover it.

Gonçalo’s magic not-withstanding, no major new results from sequencing were announced in this session. This is a shame, but in a way, it also made the session far more interesting, as the speakers had to talk about all the original thinking that went into their experiments, rather than just celebrating their dozens of newly associated disease loci.