Tutorial

This page provides a tutorial for TWAS fine-mapping using FABIO. Before runing the example code, make sure that the FABIO software is installed successfully. For instructions on installation, please see the Installation section.

FABIO #

The example data for FABIO tutorial can be accessed following this page. Here are the details about the input data formats and how to run FABIO.

1.1 Formats of input files for FABIO #

• Predicted GReX: We require the predicted GReX of the TWAS cohort built up using standard softwares like PredXican or BSLMM. The input GReX is expected to be a .txt or .txt.gz file, which has gene names as the first column, with the following columns of preicted GReX at individual-level. Each following column represents GReX of genes for an individual. An example file can be downloaded here.
• Binary phenotypes: We also require the observed binary phenotypes of the TWAS cohort. The input phenotypes are expected as a single-column .txt file, with 1 coding for case and 0 for control. The order of the individuals here should be consistent with the order of columns in the predicted GReX file. An example file can be downloaded here.

1.2 Preperation of the predicted GReX file #

FABIO provides a function to help generate the predicted GReX file in the FABIO-requested format, with the inputs:

• PLINK 1 binary files (.bed+.bim+.fam) of genotypes sorted by chromosome. Please refer to this webpage for the details of the format.
• Pre-trained eQTL weights first sorted by chromosome, then grouped by gene. The eQTL weights we used in the manuscript in the preferred file structures can be downloaded here. These weights were trained by BSLMM using GEUVADIS data.

This helper function can be applied like this:

library(FABIO)

chr <- 22
geno_dir <- "/path/to/plink/files"
weight_dir <- './GEUVADIS_BSLMM_weights/chr22'
# The eQTL weight files can be customized, but should be organized in the same format and file structures as our example files.
prepGReX(chr, geno_dir, weight_dir)
# The results will be saved as a file named "grex_for_fabio.txt.gz", in the format illustrated in the section 1.1 above.

2. Running FABIO #

The TWAS fine-mapping can be performed using the following scripts with our example data:

library(FABIO)

grex <- data.table::fread('./example_grex.txt.gz')
pheno <- scan('./example_pheno.txt',numeric())
fabio(grex, pheno, beta_a=0, beta_b=0, w_step=100, s_step=1000)
# The output results will be saved as a file named "FABIO_out.csv"
# w_step and s_step are set to be small for a shorter running time as an example here

The inputs are:

• grex: the predicted GReX
• pheno: the TWAS phenotype vector
• beta_a: alpha of the prior beta distribution on pi, both beta_a and beta_b are 0 by default, leading to the default uniform prior on log pi
• beta_b: beta of the prior beta distribution on pi, both beta_a and beta_b are 0 by default, leading to the default uniform prior on log pi
• w-step: the number of warm-up steps in MCMC, default = 6000
• s-step: the number of sampling steps in MCMC, default = 20000

3. FABIO output #

FABIO will output a summary table with three columns, and save it as a .csv file:

• Gene: name of each input gene; the genes are ordered decreasingly by PIP
• PIP: corresponding PIP of each input gene
• estFDR: estimated FDR using that gene as cutoff