PROMISED

PROMISED: Prediction of Renal Outcome using Multiomics and Integrated Statistical Evaluation of Delayed Graft Function in kidney transplant recipients.

Installation and Dependencies

This repository is tested under the following system settings:

• Python 3.7.9 (is recommended to create a venv)

Clone this repository from Github

git clone https://github.com/ginTom/PROMISED.git

CTGAN

• Install Python dependencies for CTGAN project

pip install -r CTGAN_requirements.txt

MOGONET

Clone MOGONET repository from Github and install dependencies

git clone https://github.com/txWang/MOGONET.git
pip install -r MOGONET_requirements.txt

mixOmics

Install the mixOmics R package from Bioconductor; you may need to install the latest R version and the latest BiocManager package installed.

## install BiocManager if not installed 
if (!requireNamespace("BiocManager", quietly = TRUE)) {
  install.packages("BiocManager")}
## install mixOmics 
BiocManager::install('mixOmics')

SNP features selection

A logistic regression (LR) model with recursive feature elimination (RFE) was fitted on SNPs data to select the first 100 discriminant features using sklearn python package

import pandas as pd
from sklearn.feature_selection import RFE
from sklearn.linear_model import LogisticRegression
from sklearn.feature_selection import f_classif

snp_path = 'path/to/snp/data'
snp_data = pd.read_csv(snp_path)
snp_label = snp_data['class']
snp_data.drop(columns=["class"], inplace=True)

model = LogisticRegression()
rfe = RFE(model, n_features_to_select=100)
fit = rfe.fit(snp_data, snp_label)

snp_features = snp_data.columns.to_numpy()[fit.support_]

Prepare training data

Launch CTGAN_main.py to generate and obtain training data for each omic:

python CTGAN_main.py

• label_tr.csv: labels for trainin set
• label_te.csv: labels for test set
• {1,2}_featname.csv: feature names for each omic
• {1,2}_tr.csv: traing data
• {1,2}_te.csv: test data

MOGONET

Create folder with all training data; Customize data_folder and view_list in MOGONET/main_mogonet.py

Biomarkers identification

Customize data_folder and view_list in MOGONET/main_biomarker.py

mixOmics

Launch DGF_mixomics.R to get all results from DIABLO model