What can be considered as novelty in GWAS?

Question

[statfa]

Hi,

I am asked to conduct a GWAS and to have some novelty in the study too. 1200 cases have a cardiovascular disease and 1800 controls have other types of diseases (such as diabetes, cancers, etc.). Hundereds of thousands of SNPs have been genotyped. What can be considered as novelty when GWAS can only examine the association between genetic variants (such as SNPs or CNVs) and a trait (here it is cardiovascular disease)? There are many studies conducted on this subject and I don't know what more can someone do to be considered novel. They tell me to read recent articles but I don't know what to look for more than applying the current tools on the data such as plink. I am a biostatistician, by the way, and I guess that is up to them (they are geneticists) to bring novelty not me because my job as a biostatistician is just to find a statistical method to solve a question not to find a question on behalf of them. Would Someone please give me a hint?

Thanks a lot

[Ram]

You are correct. It is not your job to strangle the data until you force a "novel" hypothesis out of it. What you're being asked to do is called a fishing expedition, and it rarely gets anywhere productive.

[Denise CS]

Have you thought in discussing this with "them"? Remind your job role to them?

I understand "their" ("our"?) need to publish. It's our PhD thesis. It's for our grant. It's for our tenure. It's for our next job.

The "publish or perish" is still on. We can rarely publish something that is not novel (whether it's an idea, method or a new association with cardiovascular disease). And if you do, it will be in journals that are less impactful (whatever our definition of impact here). There is the fear of being "scooped", and this fear is about finding novelty. And the pressure is to publish novel and positive results.

What can be novel? We will never know until we perform the experiment, analyses, until we read up, until we discuss with our peers.

Perhaps your 1200 cases are from populations less studied so far? It may suggest new loci? Do you agree with the statistical methods described/published so far?

Can you explore an idea of a new GWAS? A post-GWAS area? Non genomewide association studies?

Answer 1

[i.sudbery]

All you can do with new data is see if new SNPs are identified by it ... Nothing more.

What you might be able to do is add your patients as a metaanalysis to existing studies and apply to the meta-analysis that havn't been applied before. Perhaps they are new techniques for correcting for population structure, or new ways of look at the variance explained by the variants.