LD in LocusZoom Interactive Plot

Question

[acererak]

I am new to using LocusZoom and I am specifically using the my.locuszoom.org version of the tool.

I am trying to display my LocusZoom plots with the nearby hits colored by the r^2 coefficient, like I see is possible in some of the public studies that are posted to the site. Currently all of my "dots" are colored grey for "no r^2 data"

Is this capability possible using the my.locuszoom.org version? Or is it only possible with the standalone version? Do I need a column for LD in the input GWAS file I am uploading? I didn't see an option to do so in the uploading tool. Thanks!

Answer 1

[acererak]

So I got some answers from Locus Zoom I'd like to share!

"The new site doesn't yet support user-provided LD values, but it can automatically fill in LD information based on the 1000G reference panel (for your choice of several sub populations). The two caveats are:

1. You need a full variant specification: chromosome, position, reference and alt allele.
2. The 1000G reference panel is a bit old, and some new datasets (especially highly imputed ones) have variants not known to that panel. Since LD is calculated relative to the most significant SNP in the region, you would see no LD information relative to an unknown variant. (and all points would appear grey)

You can change the reference variant interactively on your plot by clicking on any SNP in the plot, then selecting "make LD reference". Sometimes that can shift the calculation to use a variant that is present in 1000G and colors will appear. We would like to use newer LD information, but there are some data access issues involved. Alas, I can't offer a timeline, beyond saying that it is a subject of ongoing interest to us."

For my case, since I am using an imputed dataset I just didn't have a hit that had an 1000 Genomes reference.

Answer 2

[abought]

Thanks for sharing this information! Since our conversation, we have improved the panel used for LD in some cases.

For GRCh38 datasets, plots will now default to using a new and improved version of the 1000G LD reference panel that is based on deep whole-genome sequencing. No further action is needed to take advantage of this feature. Feedback on the new feature is always welcome. :)

Datasets in build GRCh37 will continue to use the older 1000G reference panel; at present we have no plans to translate the new LD panel to older genome builds.