[GET-Editors] LOD scores

[Madeleine Price Ball]

I have been following the pop-up... so I say, do that for now. Just
make sure to note the LOD score in the comment so it's easy to change
later.

The guide is inconsistent because at some point I was fussing over the
inconsistency between LOD and p-values used for case/control. But the
values in the guide are still not consistent with the the case/control
p-values, so I wouldn't worry about following it right now.

If I understand it correctly, you can roughly map LOD to p-value:
LOD (log odds) is often: log (1 / p-value)  (the first 1 is if you if
the variant perfectly explains the disease)
then, p-value = 1 / (10**LOD)

The case/control p-values are:
1 pt: p <= 0.1
2 pts: p <= 0.05
3 pts: p <= 0.025
4 pts: p <= 0.01
5 pts: p <= 0.0001

If we made LOD consistent, it'd be:
1 pt: LOD >= 1.0
2 pt: LOD >= 1.3
3 pt: LOD >= 1.6
4 pt: LOD >= 2
5 pt: LOD >= 4

For reference, the current pop-up says: (0.5, 1, 1.5, 3, 5) and the
current guide says (1.0, 1.3, 1.5, 3, 5).

  - Madeleine

On Wed, Nov 24, 2010 at 3:31 PM, Abraham Rosenbaum <rosenbaum4 at gmail.com> wrote:
> The how-to list has a score of 1 for LOD 1 or above, and the floating
> helper-friend has a score of 1 for a LOD of 0.5 or above. Which one are we
> following?
> Thanks,
> Abraham
> _______________________________________________
> GET-Editors mailing list
> GET-Editors at lists.freelogy.org
> http://lists.freelogy.org/mailman/listinfo/get-editors
>
>