4 ms·

Follow-ups to see what other phenotypes are associated with GWAS hits have occupied a meaningful fraction of the mindshare of statistical genetics over the past

Follow-ups to see what other phenotypes are associated with GWAS hits have occupied a meaningful fraction of the mindshare of statistical genetics over the past few years: GREML, BOLT-LMM’s REML, and MTAG are examples of tools for evaluation and exploitation of the genetic correlation between traits. I would add that PheWAS is often performed with specific genetic instruments of interest, and can be performed with aggregate measures such as a genetic risk score. This lets you see association across a slew of traits at once.

subroutine 8y ago
I was aware of the Broad tools, but point taken. There's also databases like GeneCards (e.g. bit.ly/APOE_genecards ) or just the NIH genome viewer: (e.g. bit.ly/APOE_NCBI ) which have SNP accession numbers, and annotations for disease-linked variants. And it's good we are sharing this info. But we are far from having high-powered GWAS for all of human's traits and diseases. And often what we do know is from a single demographic (fwiw mostly caucasian, not chinese). How gene variants influence outcomes in different genetic backgrounds, or within certain environments (think developmental nutrition in africa vs asia) is typically a big questionmark. So IMO performing CRISPR edits to mitigate a problem that was completely avoidable by other means was unethical at best.
- carbocation 8y ago
  We are in violent agreement about the ethics. For UK Biobank, there is a GWAS of ~everything: http://www.nealelab.is/uk-biobank/ http://www.nealelab.is/uk-biobank/ The lack of diversity in genetic studies is a huge problem. This one isn't just moral, either: we're ignoring the richness of allelic diversity that is found in people throughout the world (particularly in Africa, which has the most genetic diversity because of its pre-bottleneck status). There is an awareness and I think you'll see big changes here in the next few years.

subroutine 8y ago

I was aware of the Broad tools, but point taken. There's also databases like GeneCards (e.g. bit.ly/APOE_genecards ) or just the NIH genome viewer: (e.g. bit.ly/APOE_NCBI ) which have SNP accession numbers, and annotations for disease-linked variants. And it's good we are sharing this info. But we are far from having high-powered GWAS for all of human's traits and diseases. And often what we do know is from a single demographic (fwiw mostly caucasian, not chinese). How gene variants influence outcomes in different genetic backgrounds, or within certain environments (think developmental nutrition in africa vs asia) is typically a big questionmark. So IMO performing CRISPR edits to mitigate a problem that was completely avoidable by other means was unethical at best.

carbocation 8y ago
We are in violent agreement about the ethics. For UK Biobank, there is a GWAS of ~everything: http://www.nealelab.is/uk-biobank/ http://www.nealelab.is/uk-biobank/ The lack of diversity in genetic studies is a huge problem. This one isn't just moral, either: we're ignoring the richness of allelic diversity that is found in people throughout the world (particularly in Africa, which has the most genetic diversity because of its pre-bottleneck status). There is an awareness and I think you'll see big changes here in the next few years.