You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Also really important because sometimes credible sets span quite a long distance, so the normal LD window might not be large enough. Can be seen in http://results.finngen.fi/region/C3_BREAST_EXALLC/9:107191059-107691059, where the 4th susie credible set spans almost 3MB, and the variant with largest PIP is over 2 MB from the variant with largest position.
I looked at the possibilities for the different APIs, and here's my impressions:
plink has pairwise API for only 1 variant pair at a time, which is ridiculous. The calculation time for one pair is dominated by the need to load the allele frequencies into the computer memory. It might be better to just use the range-based LD calculation, and adjust the range based on the variants that were supplied to the function.
LD server has only LD for a range. It can be used as long as the distance between two variants is less than 2.5MB. The output then needs to be filtered down to the correct pairs, but that's quite simple.
Seems like for both of those LD backends, it's best to use the range-based calculation and then filter out the desired results.
pairwise LD api would be very good for getting credible set LD, as discussed in #103.
In PLINK, I can use the --r2 flag with --ld-snp-list flag.
In online api, I'll have to fetch appropriate range and filter the results, but it's not too much of a hassle.
The text was updated successfully, but these errors were encountered: