We propose a novel Bayesian model, CARMA (CAusal Robust mapping method in Meta-Analysis studies), for fine-mapping in order to identify putative causal variants at GWAS loci. The main features of CARMA are
Modeling jointly summary statistics and high-dimensional functional annotations
Introducing a novel Bayesian hypothesis testing approach to account for discrepancies between summary statistics and LD from external reference panels
flexible specification of the prior distribution of effect sizes including a heavy-tail Cauchy distribution that is more adaptive to the possibly different signal-to-noise ratios across loci
Vignettes can be found at [https://github.com/ZikunY/CARMA/blob/master/CARMA_demo.pdf].
The manual can be found at [https://github.com/ZikunY/CARMA/blob/master/CARMA.pdf].
PLINK is a popular and useful toolset used for generating LD matrix from either VCF or BED file. It is important to notice that the command ``--r'' should be specified to generate $r$ LD, instead of $r^2$ LD as the $r^2$ LD does not account for the directions (sign) of the summary statistics. Additionally, maintaining consistency in the coding order of testing alleles between summary statistics and the LD from the reference panels is also crucial for fine-mapping analysis. The following command is an example of generating LD matrix with PLINK while ensuring the coding order of testing alleles remain consistent.
./plink --bfile bed.file --a1-allele SNP.txt 2 1 '#' --r --matrix --out LDThe ``SNP.txt'' file includes the information of SNP names and ref alleles from the GWAS summary statistics. Please check plink manuals for LD calculations https://www.cog-genomics.org/plink/1.9/ld#r
devtools::install_github("ZikunY/CARMA")CARMA requires the Intel MLK library, therefore please specify the pathway for the Intel MLK library. The Intel MLK library can be installed by the following lines
conda install -c anaconda mkl
LD_PRELOAD=pathto/anaconda3/lib/libmkl_rt.so RImportant Note
The pathway for the Intel MLK library and related libraries may differ depending on the operating system. Users should verify the correct pathway on their system and modify the LD_PRELOAD accordingly.
If users encounter issues while installing CARMA on an M1 Mac, it may be due to the failure to locate the GSL library or other libraries during the installation process. The problem could be solved by modifying the ‘PKG_CPPFLAGS’ environment variable in the /.R/Makevars file’. This file should be created if it doesn’t exist. Open the/.R/Makevars’ file in a text editor, and add the following line to set the PKG_CPPFLAGS environment variable:
PKG_CPPFLAGS=-I/opt/homebrew/Cellar/gsl/2.7.1/include/The pathway above corresponds to the GSL header files’ location for GSL version 2.7.1. Depending on the user’s system, this pathway may vary.
Setting PKG_LIBS
Next, add the necessary lines to the ~/.R/Makevars file to specify the GSL library’s location and other required libraries:
PKG_LIBS=-L/opt/homebrew/lib -lgsl -lgslcblas -lmThe above line tells the linker (ld) to search for the GSL library (libgsl) and GSL CBLAS (libgslcblas) in the Homebrew installation directory (/opt/homebrew/lib). The -lm flag links the math library.
Yang, Z., Wang C., Khan A, Vardarajan B, Mayeux R, Kiryluk D, Ionita-Laza I,
CARMA: Novel Bayesian model for fine-mapping in meta-analysis studies