Gee, what are GWAS?

Perhaps you've been reading in the news lately about genes "for" autism, obesity, cancer, sudden heart attack . . . . . . and on and on.

What are all these reports about? 

Most of the reports you see are based on genome-wide association studies, or GWAS. You pronounce this acronym: gee wahs.

The genome, by the way, is the full set of DNA (the inherited material) in a human, mouse, yeast, or whatever is being studied. In most interpretations, genomics--the science of genomes--includes the study of environmental as well as genetic information. However, GWAS are usually studies of the DNA of large populations of people with disease X compared with 1) people who don't obviously have the disease or with 2) some sort of DNA standard.

And right away, we hit the first issues of GWAS studies:

  • How is the disease identified or defined? For example, is the diabetes Type I or II? Does autism include autism-related syndromes? Is the lung cancer non-small cell, mesothelioma, or something else?
  • People who don't obviously have a disease might develop it at a later time. So, your "control" population may be tainted.  
  • Who decides what the DNA standard is? It could be the human genome (a compilation of several different DNAs), one individual's DNA, or some other source.

Next, what is already known about the genetic components of this disease? We know, for example, that Huntington's Disease is inherited in a simple dominant fashion, so that an adult who has one parent with the disease has a 50% chance of inheriting the disease--not 25% or less, which would be the case for a recessive gene. Meanwhile, the "breast cancer genes"--BRCA1 and BRCA2--are often used to indicate the possibility of developing breast cancer, but they represent only 5% of all breast cancer cases and many more genes are now known. Complex--or multifactorial--diseases most likely have many genetic components and environmental factors to consider.

A key problem is this: Association does not mean causation. So, just because storks have been associated with baby births, doesn't mean they bring the babies. And, because a DNA piece is associated with a disease doesn't mean it causes the disease.

Most of the GWAS associations are based on SNP studies, that is, studies of single nucleotide changes (polymorphisms) in a small piece of DNA. And, most known SNPs are found in nonfunctional portions of DNA--areas not responsible for the production of a known protein. 

Finally, are you going to trust statistical analyses--where large numbers can easily lead to "statistically significant", but clinically meaningless, results--to decide whether you may develop disease or condition X?

I hope not. 

By the way, for most complex conditions, such as diabetes or cancer, as many as 400 genes may need to be identified before a clear "causation" might be identified. Seems we have a long way to go. 

Marie Godfrey, PhD

| | | | | mgodfrey39's blog