TWAS

Preprocessing of TWAS Data:

Reports on all preprocessing steps can be seen in the respective subfolders in the synapse project here:

Synapse Data Project

All Expression data is re-processed to correct for diagnosis with the following scripts

• code/MSBB_RegressDiagnosis_4_TWAS.Rmd
• code/Mayo_RegressDiagnosis_4_TWAS.Rmd
• code/ROSMAP_RegressDiagnosis_4_TWAS.Rmd

All Variant data processing and Ancestry Culstering is preformed with:

• code/Ancestry_PCA.Rmd

Harmonization of Sample names and RNA-Seq Dataframe Finalization can be found here:

• code/ProcessData.Rmd

Running Code

Log into EC2 instance

 #Pull Repo
 git clone https://github.com/jgockley62/TWAS.git
 
 #Build Image from Dockerfile
 docker image build --build-arg USER_ID=$(id -u ${USER}) --build-arg GROUP_ID=$(id -g ${USER}) -t twas /home/${USER}/TWAS/
 
 #Open Docker container and RInstance
 docker run -v "/home/${USER}/TWAS/:/home/${USER}/TWAS/" -e USER=${USER} -e PASSWORD=<PASSWORD> -d -p 8787:8787 <IMAGE_ID>

Now loginto RStudio Instance https://<AWS_Instance_IP>:8787

setwd("~/TWAS/code/")

#Sample Code Run Command:

synapseclient <- reticulate::import("synapseclient")
syn_temp <- synapseclient$Synapse()
syn_temp$login()

library(data.table)

source("~/TWAS/utilityFunctions/knitfile2synapseClient.R")
source("~/TWAS/utilityFunctions/hook_synapseMdSyntax_plot.R")

createAndKnitToFolderEntityClient(file = "ProcessData.Rmd",
                                          parentId ="syn18936948",
                                          folderName = 'Processed_TWAS_Training_Data')

Train Weights on AWS Instance

docker build -t train /home/${USER}/TWAS/TWAS_ModelTraining/

screen

docker run -it <Image_ID> /bin/bash

cd fusion_twas-master

synapse get syn20835005
synapse get syn20835007
synapse get syn20834996

synapse get syn20857313
synapse get syn20857312

git clone https://github.com/jgockley62/TWAS.git
cp TWAS/code/WorksMultiThreaded.py .
cp TWAS/code/FUSION.compute_weights_Parallel_V2.R .

awk '{ print $1"\t"$2 }' All_CEU_ToTrainForTWAS.fam > Total_IDs.txt

#Adjust values PRE_GEXP, INPT, cores

python WorksMultiThreaded.py