Show correlation between phenotypes in PheWAS plots

[pjvandehaar]

1. Compute the (spearman?) correlation matrix between all phenotypes (ie, ignoring DNA/associations).
2. on a PheWAS plot, make the highest association the "reference".
   • this should be changeable with a toolbar button "select reference phenotype"
3. color other traits according to their correlation with the reference.
4. allow conditioning on a phenotype.

That is, we should support "LD" among phenotypes much like LD among variants. This would help with interpreting whether peaks below the top are independent or just due to correlated traits.

[abecasis]

This is a cool idea. I wonder what it would look like.

G

Sent from my iPhone

On Mar 24, 2017, at 6:16 PM, Peter VandeHaar [email protected] wrote:

Compute the correlation matrix between all phenotypes (ie, ignoring DNA/associations). on a PheWAS plot, make the highest association the "reference". this should be changeable with a toolbar button "select reference phenotype" color other traits according to their correlation with the reference. allow conditioning on a phenotype. That is, we should support "LD" among phenotypes much like LD among variants. This would help with interpreting whether peaks below the top are independent or just due to correlated traits.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

[pjvandehaar]

credit to @sgagliano for bringing it up.

[sgagliano]

Let's make this a high priority to aid in the interpretation of either pleiotropic or redundant signals the PheWAS plots.