ADNI sex differences analysis

This is the repository of the Sex Differences in the Metabolome of Alzheimer's Disease Progression article.

Authors: Tomás González Zarzar, Brian Lee, Rory Coughlin, Dokyoon Kim, Li Shen and Molly A. Hall.

The purpose is to detect sex differences in metabolite-phenotype associations.

Clone repository

To clone the respository, on your terminal type:

git clone https://github.com/tomszar/ADNI_project.git

Then, enter the respository and follow the next instructions

cd ADNI_project

Environment setup

The repository provides an environment.yml file to use with conda

First, install anaconda on your local computer or user server account following the appropriate directions, using the corresponding installer to your operative system. Next, on the terminal, in the root of this repository, install the conda environment by running:

conda env create -f environment.yml

If, for any reason, the installation through the environment file fails, you can install the environment manually, by running the following:

conda config --add channels r
conda config --add channels bioconda
conda config --add channels conda-forge
conda create --name adni_project python=3.9 pandas numpy scipy jupyterlab statsmodels scikit-learn pingouin r-base r-wgcna r-dplyr
conda activate adni_project
pip install clarite

Replicate the analysis

Depending on whether you are replicating the analysis on your local machine or sending the job to a cluster, you can run either run_local.sh or run_cluster.sh

On cluster

On your server terminal, run the following command:

qsub run_cluster.sh

The parameters used in the script are the ones used in the Penn State Roar server. Depending on the system you will need to modify, remove, or add parameters to the script. The script also contains the copying of the needed ADNI data sets into the data folder; you can either modify the paths of the data sets in the ADNI_data_files.txt file to match yours, or manually copy the files, and comment the lines in the run_cluster.sh script. In the next section there is a brief description on the ADNI data sets needed.

On local machine

Before running the pipeline, you will need to create the data folder and drop the files needed from ADNI. On the terminal, type

mkdir data/

Copy the following files from ADNI in the data folder

• p180 data files: ADMCDUKEP180UPLC_01_15_16.csv, ADMCDUKEP180FIA_01_15_16.csv, ADMCDUKEP180UPLCADNI2GO.csv, ADMCDUKEP180FIAADNI2GO.csv, ADMCDUKEP180FIAADNI2GO_DICT.csv, ADMCDUKEP180UPLCADNI2GO_DICT.csv, 4610 FIA p180 Data.xlsx, 4610 UPLC p180 Data.xlsx.
• nmr nightingale files: ADNINIGHTINGALE2.csv, ADNINIGHTINGALE2_DICT.csv.
• fasting information: BIOMARK.csv.
• LOD values: P180FIALODvalues_ADNI1.csv, P180FIALODvalues_ADNI2GO.csv, P180UPLCLODvalues_ADNI1.csv, P180UPLCLODvalues_ADNI2GO.csv.
• QT-pad: ADNI_adnimerge_20170629_QT-freeze.csv.
• Drug classes: ADMCPATIENTDRUGCLASSES_20170512.csv

Next, on the terminal, type:

conda activate adni_project
bash run_local.sh > run_local.log