Comments on: Are synthetic associations a man-made phenomenon? https://genomesunzipped.org/2011/01/are-synthetic-associations-a-man-made-phenomenon.php public personal genomics Sat, 27 Jun 2015 11:37:55 +0000 hourly 1 http://wordpress.org/?v=4.2.5 By: shane maehttps://genomesunzipped.org/2011/01/are-synthetic-associations-a-man-made-phenomenon.php#comment-321787 Mon, 17 Sep 2012 12:56:29 +0000 https://genomesunzipped.org/?p=2449#comment-321787 Jeff, I meant to say hi at the DDD research forum today; loving this blog. I confess to feeling heavily out of my depth when discussing these problems, but isn’t the main problem not that the majority of GWAS have been successfully replicated in different populations, but only those that *have* been successfully replicated have been shown to be replicable in different populations? Many results remain unreplicated and may be type 1 errors (I don’t have the figures, and I may be totally over-egging the pudding here!).

That said, supposing synthetic association is common (and probabilistically this strikes me as unlikely), isn’t it still the case that the data erroneously lead you to the right gene, rather than the wrong gene? In which case, does it really matter? And when you sequence the gene in multiple cases (SNP typing being *so* last decade), all will be revealed?

OK, I get the point that we’re still chasing the missing heritability, but the only way we’re likely to find it is by using the *non*missing heritability to unpick the biology, and then design better candidate gene studies (likely by in silico analyses of exome/genome data)?

Sorry if these questions are a tad random and ignorant, but I have a hard time bending the old brain around these complex statistical issues, yet I think they’re important. Thanks for flagging these up – that’s my night-time reading sorted…

Cheers,
shane mae

]]> By: Shane McKeehttps://genomesunzipped.org/2011/01/are-synthetic-associations-a-man-made-phenomenon.php#comment-31487 Wed, 19 Jan 2011 21:58:56 +0000 https://genomesunzipped.org/?p=2449#comment-31487 Jeff, I meant to say hi at the DDD research forum today; loving this blog. I confess to feeling heavily out of my depth when discussing these problems, but isn’t the main problem not that the majority of GWAS have been successfully replicated in different populations, but only those that *have* been successfully replicated have been shown to be replicable in different populations? Many results remain unreplicated and may be type 1 errors (I don’t have the figures, and I may be totally over-egging the pudding here!).

That said, supposing synthetic association is common (and probabilistically this strikes me as unlikely), isn’t it still the case that the data erroneously lead you to the right gene, rather than the wrong gene? In which case, does it really matter? And when you sequence the gene in multiple cases (SNP typing being *so* last decade), all will be revealed?

OK, I get the point that we’re still chasing the missing heritability, but the only way we’re likely to find it is by using the *non*missing heritability to unpick the biology, and then design better candidate gene studies (likely by in silico analyses of exome/genome data)?

Sorry if these questions are a tad random and ignorant, but I have a hard time bending the old brain around these complex statistical issues, yet I think they’re important. Thanks for flagging these up – that’s my night-time reading sorted…

Cheers,
-Shane

]]>