Candidate gene and GWAS are both essentially gene association studies and are subject to the same artefacts of spurious association findings. GWAS rely on indirect association to locate a disease-causing variant. Direct candidate gene analysis relies on an a priori hypothesis to identify a disease-causing variant by direct sequencing. The future probably will consist of whole genome sequencing, capturing both data, with adequate filtering of the huge amount of data obtained in order to extract the signal from the noise.

I think that NGS are dropping down genome sequencing costs and times. Mapping is now easy and fast as well as SNPs recruitment. These data are always more easy to obtain and to compare enlarging databases at the same time. On the other hand the studies could be focused on specific genes of interest. My view is that wide genomes analysis jointly with database curation should go at the same time with less laborious methods such as PCR although I think that now-a-day there are already the bases (or we are very close) for thinking to whole genome sequencing as a "less-laborious-method" than PCR.

Due to the continuing drop in sequencing costs with improved technology and single molecule sequencing, the goal should be a complete genome sequence for every person at a cost of around $200, which is comparable to other lab tests and should soon be within reach.

Everything depend on wich organism you are working... For this question I read a review some years ago... you should google it...

So If you use candidate gene, you should have a good background of what gene could be... If you work with genome wide, you may have relation that are artifactual...

Being working on an orphan organism, I use genome wide markers... I have no idea if I will at some point have to work at the gene scale... I even use the genomic scale (Gish fish etc... ) that is also informative...

Andy Grove said in YouTube 5 years ago http://www.youtube.com/watch?v=H39cYW0xgbM at min 4 sec 10. "The holy grail of all of this it would be a biological measure of progression that is objective, repeatable and rock-solid as a standard.

We don't have that". In 2007, we did not have an affordable whole genome analysis - and are still in the early days of turning PoC-s of Fractal Defect mining into cancer diagnostics & therapy by fractal analysis. We should not have lost Steve Jobs. 2012 is a turning-point. - Pellionisz

Candidate gene and GWAS are both essentially gene association studies and are subject to the same artefacts of spurious association findings. GWAS rely on indirect association to locate a disease-causing variant. Direct candidate gene analysis relies on an a priori hypothesis to identify a disease-causing variant by direct sequencing. The future probably will consist of whole genome sequencing, capturing both data, with adequate filtering of the huge amount of data obtained in order to extract the signal from the noise.

Both candidate gene and GWAS are important.

First clear understanding of genetic architecture of the disease is very much needed. Then only we can interpret the results, followed by choosing appropriate analysis.

GWAS paved a way for the exploitation of available database to identify the factors for various diseases. Variety of mutations identified in all diseases with respect to population wise, geographical area wise.

It is important to note that what kind of data we are searching from the database. While analyzing the data by using various kinds of available software’s, it is necessary to keep in mind what we are exactly looking and interpret from the outcome. Each, Software is designed based on certain algorithms suit for particular kind of outcome. Basing on that we can interpret the results but the result could not be generalized which may lead to wrong conclusion.

In the context of animal breeding to predict production or disease vulnerability that while genome wide gives some information, a candidate gene approach to actually create something useful (predicative). However, the majority candidate gene tests that have gained commercial traction effectively Identify "mutations" that have create a drastic disease like effect. Subtle single gene effects appear to be highly breed specific (so gene X 'causes' Y, but this can only be reliably observed in Friesian cows). This breed specific effect is also seen in humans with the latest whole genome work indicating that only a tiny number of expressed genes are linked heart disease, and most of these only occur in a particular racial background.

So in answer to the original question-candidate genes are the only way to produce a usable product right now, but this is generally limited testing for drastic effects such as disastrous mutations. Whole genome analysis is still nowhere near good enough or cheap enough to deliver a useable product. It is limited by the confounding effects of familial links (both genetic and environmental) and racial background creating genetic bias and fundamentally whole genome analysis cannot answer the question "what is important?” in a genome.

As to what happens in the future? I’m not seeing whole genome analysis delivering reliable predicative power any time soon. The more complete and complex genome analyses we do, the less and less we appear to be able to predict about phenotype. This implies that either our previous analyses were flawed OR gene to gene to protein to environment interactions are so complex that most effects are to subtle to measure against the background of the entire genome).

At the end of the day phenotype is still king for measuring of genetic performance.

From the animal world, yes a genetic risk scoring is possible, BUT it depends on the trait and if it is quantitative and how heritable it is (so how big is the genetic component of genes+ environment =phenotype)

E.g. Look at 2 sheep diseases, facial eczema caused by fungal toxins in their grass and footrot. In facial eczema, highly resistant sheep can be found and it is possible to quantitatively assess resistance by exposing sheep to toxin. this allowed very good genetic studies to be undertaken. In footrot, there is no quantitative range of disease-you either have it or not. Furthermore footrot genetics don't manifest as resistant disease proof animals, rather risk of disease vulnerability is identified which in turn means genetic testing is used in sire selection to make sure ram is not carrying a "compromised" gene.

So is whole genome sequencing useful to approach these diseases??

In the case of facial eczema, maybe, the genetic influence on the disease outcome appears high, will be controlled by multiple genes which should be additive to the phenotype.

Footrot is much harder, we know the MHC genes are critical to the immune response, however these genes are hyper variable (100's of variants described) and the environmental influence on footrot is very large making it very difficult to conduct any kind of consistent challenge to measure the genetic response to disease. Worse, the current whole genome testing technologies (SNP chips) cannot resolve all the variants in MHC (or even get near). So even though we know MHC is important, that some variants are highly predicative of relative risk we cannot use a whole genome approach to analyse footrot vulnerability.

To highlight the difficulties just consider trying to recruit 1000 animals carrying a common (e.g. 5% frequency) combination of MHC genes. You need to screen 20 000 animals. This would be one of the few ways of controlling for the confounding effect of a MHC gene allowing you to study the rest of genome. If the combination of MHC genes was relatively uncommon, say with frequency of 0.5% (very possible e.g. 2 genes one at a 10% frequency and the other at 5% frequency), you need to screen 400 000 animals to find 1000 carrying the combination of genes you want to study.

This math becomes much worse if you want to look at relatively uncommon genes or combinations of 2 or 3 genes.

Note in latest genetic study of rare genetics in humans, 5700 people were studied and all their expressed genes sequenced, this identified 1.2 million gene variants, of these 1.1 million were rare variants, some so unusual that only 1 person in the study carried them.

This suggests to me, meaningful whole genome analysis could be several orders of magnitude more difficult than we previously thought….At best it can tell us where to look in detail in the future.

Always agreing , Grant. It's the reason why most of human studies, to determine genetic risks of disease, used in the past case/control study with candidate-gene approach, but with the statistical notion of selection of a sample from a general population (even with 1000 or 10.000 subjects) and extrapolation to the general population; and in general you study the repartition of a genotype (in this case, often in families), or polymorphisms for "mutations" found in all people, but with difference between wild or homo or hetero mutated subjects.. And you conclude (or not) to an associationbetween a polymorphism and a disease, with no casualty relation becaus you don't study at the same time all the factors that could be linked to the disease.

And the new studies with transciptome don't resolve the problem, even with thousands of SNP studied at the same time on thousands of subjects. And the statistical analysis is more difficult in this last case because of the technic (you select only as positive the SNP detected with a sufficient statistical association significance). And using these methods, most of researchers have to repeat their GWA to confirm results (cost much more prohibitive !).

Regards

Didier

I think they are mutually complementary in gene function research. It represents forward genetics and reverse genetics respectively.In fact, GWAS can help us identify the variants related to phenotype or other traits, but to some extent, it is just probability. Then, its results need to be confirmed by other methods. So, to deeply explore the gene's function and their difference in diverse individuals, both of them are indispensable.