Table of contents

• Project description
• Setup
• Run

Project description

Build reproduciple pipleline of comparative analysis of multi-omics biomarker discovery project. Implemented the following tools in nextflow, which is a workflow manager.

• MOFA (2018, Mol. Sys. Biol.)
  MOFA is originally designed for unsupervised integration of multi-omics data. As biomarker discovery problem is feature selection probelm in supervised task, we fed a logistic regression model with patient representation generated by MOFA to measure the predictive power of biomarkers.
• iDRW (2021, Bioinformatics)
  iDRW is supervised multi-omics integration method that represents a patient with a pathway activity vector. To avoid information leakage, we modified iDRW (Code available at ./iDRW/R/get.iDRWP.R)and additionally implemented iDRW_test (Code available at ./iDRW/R/get.idRWP_test.R).
• DIABLO (2019, Bioinformatics)
  Multi-omics biomarker discovery tool based on canonical correlation analysis.
• MOGONET (2021, Nat. Comm.)
  Multi-omics biomarker discovery tool based on graph neural network on the basis of multi-omics patient similarity network.

Setup

Build docker image

# Pull base image from docker hub
docker pull dabinjeong/cuda:10.1-cudnn7-devel-ubuntu18.04

# Build docker image
docker build --tag biomarker_comparison:0.1.1 .

Install workflow manager: Nextflow

conda create -n biomarker python=3.9
conda activate biomarker
conda install -c bioconda nextflow=21.04.0

Run

nextflow run comparison.nf -c comparison.config -with-docker biomarker_comparison:0.1.1