--- TWAS ---

The following packages by the developers

are required and be unpacked. In addition, lists of genes in the three populations are made through the following scripts,

TWAS=/genetics/bin/TWAS
cd $TWAS
for pop in MET NTR YFS
do
  ls WEIGHTS_$1 | sed 's\/\\g' > $pop.lst
done

--- TWAS-pipeline ---

The pipeline is installed as follows,

git clone https://github.com/jinghuazhao/TWAS-pipeline

On our system, TWAS.sh and TWAS_get_weights.sh for TWAS and twas.sh, twas2.sh, twas2-collect.sh and twas2-1.sh for TWAS-pipeline have symbolic links under /genetics/bin and available from the $PATH environment. A Stata equivalent has been developed by Dr Jian'an Luan.

To accommodate the suggestion of p value in accordance with the Z-score in the output, pnorm.c is included which can be compiled as follows,

gcc pnorm.c -lm -o pnorm

and a call pnorm z_score yields a p value with more decimal places. However, the function is now part of twas-single.sh.

--- GNU Parallel ---

Further information is available from here.

Suppose you have a file containing GWAS summary statistics, you cna run the pipeline as follows,

twas.sh input_file

where the input_file is in tab-delimited format containing SNP_name, SNP_pos, Ref_allele, Alt_allele, Beta and SE. The output will be contained in <input file>.imp.

This assumes that ssh can access nodes in a clusters freely and in case this has not been done, a single node mode is more appropriate,

twas-single.sh input_file

These are available from the wiki page,

• An example on body bone mineral density

• An exposition with GIANT data

• Building reference panel

• TWAS using GTEx

• Epigenomewide association

• FUSION pipeline

The work is possible with an EWAS project within the MRC Epidemiology Unit, for which colleagues and collaborators have contributed.

Gusev A, et al. (2016). Integrative approaches for large-scale transcriptome-wide association studies. Nature Genetics (a copy at Harvard), 48, 245-252