Is genome wide association study meeting its end?

A paper was published in Nature last year.   The authors used 183,727 individuals to do a genome wide association study (GWAS) on human height, with 2,834,208 (imputed) loci SNP chips.



180 loci were found to be affecting human height.   Altogether, they explain 10% of  phenotypic variance of human height, or ONLY 12.5% of total height inheritable variance (h2_height=0.8).

From the point of view of statistics, the results are hard to believe.   Usually, it is called model selection to determine number (and effects) of QTLs along a chromosome.  If you have a look at  Goffinet & Mangin's paper many years ago, it is even very hard to determine the situation of 3 linked QTL per chromosome.

In a typical GWAS, people usually don't have so much funding to carry out a very large experiment as the one in this Nature paper.  The study above itself is actually a meta analysis of results from many labs around the world who have conducted some other GWAS.   They happened to have recorded the patients' heights also.

In a small experiment, one typically see several large peaks and many small peaks.   But when you zoom in the peaks, the very significant QTLs are usually gone  as they might just be some aggregate effects of many small QTLs and might even in variant association/linkage phases.

Many people believe that there are 2000+ genes affecting human height.   Some other people also estimated the number of genes that might affect common farm animal traits.  All of them suggest the lower bound of genes that affect a trait is in hundreds, or at least tens along a chromosome averagely.  The case of DGAT1 is just like a case of winning of a lottery.

We can also put the things this way.   Linkage analysis has been doing for more than 30 years.  Thought it is said that it saved the quantitative genetics, which is true. Statistics and computer sciences are now playing a even more important role today.  There were however rare responses to linkage analysis in agricultural industries.

Only 3-4 years ago, very few people knew genomic selection.  Now almost every people in animal and plant sciences is talking about it.  The industries of all kinds of farm animals and plants are investing or adopting this new technology.  Some country, like France, has totally stopped progeny tests of dairy cattle.

The world is changed.