Any guidelines to use the Haplotype Resource Consortium (HRC) as reference? (e.g., for the UK Biobank)

114 views
Skip to first unread message

Jiang

unread,
Nov 2, 2017, 9:43:07 PM11/2/17
to ldsc_users
Hi,

In the inital LD Score regression paper in Nature Genetics, the following sentence is included:

For LD Score regression, we excluded variants with EUR MAF < 1% because the LD Score standard errors for these variants were very high (...).

Later on, in the LD Hub paper in Bioinformatics, the following sentence is included:

In the future, as the ability to impute lower frequency SNPs improves we will investigate the possibility of including other SNPs in the analysis using resources like the Haplotype Resource Consortium (HRC).

I was wondering whether anyone has tried using the HRC as reference, and/or what would be the main guidelines to do so. In particular, it seems like there are many "rare" variants (e.g., MAF<1%) which are (kind of) in linkage equilibrium, so I don't know how to proceed there.

Would be great if you could provide LDSC files to use with HRC-imputed data (e.g. UK Biobank) or at least provide suggestions on how to do it.

Thanks,

Raymond Walters

unread,
Nov 3, 2017, 8:51:10 PM11/3/17
to Jiang, ldsc_users
Hi,

I’ve heard some discussion of doing a HRC-based reference, but as far as I know nobody has made one available yet (possibly related to the tighter data access restrictions for HRC vs. 1000 Genomes).

Relevant considerations for a HRC-based reference are likely to include:
1) The larger impact of the chosen assumption about the relationship between allele frequency and effect size once you get into rarer variants (e.g. from --pq-exp)
2) The change in interpretation for the quantities being estimated by LDSC if you change the range of variants included in the LD score (see e.g. section 1.7 on flavors of heritability in the supplement of the genetic correlation paper)
3) Evaluating stability of LD scores at low MAF in the HRC data
4) Revisiting the set of SNPs to use in regression (e.g. settings for munge_sumstats.py)

Cheers,
Raymond



--
You received this message because you are subscribed to the Google Groups "ldsc_users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ldsc_users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ldsc_users/c8ec2e76-9b52-435a-87dc-b3d92d622e9d%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply all
Reply to author
Forward
0 new messages