Which Tool To Use For A Full Ld Analysis?

Question

[Philipp Bayer]

Hello,

I'm currently trying to get a full LD analysis (each SNP compared against each SNP) on a chromosome of about 33,000 SNPs, but using TASSEL 3 without any special options this takes forever and blocks resources (so far 7 days and 90GB of RAM).

Does anyone have any ideas (or alternatives) to perform this type of analysis? I'm currently thinking of "settling" on only the 100 SNP sliding window approach, but it would still be nice to see the LD over the entire chromosome....

Thanks!

Answer 1

[Neilfws]

I assume this is "pairwise LD". I've not tried any of them but tools which can do that include:

• SNAP Pairwise LD (web interface, Broad Institute)
• PLINK
• See this question and answers at CrossValidated - mentions snpMatrix, LDheatmap, mapLD (R/Bioconductor)
• Is this Python code useful?
• LDhat ? C/C++ so should be fast

[Philipp Bayer]

Thanks for that! I'm currently looking into PLINK, if only it would allow tri-allelic SNPs.. Got some cleaning to do.

Cheers!

Answer 2

[Bioch'Ti]

A shorter approach would be to analyse the LD patterns at the chromosome scale at first... Moreover, you could use also LDcorSV, an [R] package that takes into account the population structure (Q) and the Kinship (K) in its estimation of r2. See Mangin et al., 2012 and http://cran.r-project.org/web/packages/LDcorSV/index.html

Best !