MultiNEP: a Multi-omics Network Enhancement framework for Prioritizing disease genes and metabolites simultanuously

MultiNEP is an improved analytical tool to prioritize disease-associated genes and metabolites simultanuously using multi-omics network with the ability to handle network imbalance. Multinep first reweight a general multi-omics network $S^0$ from database and a multi-omics similarity matrix $E$ based on disease profiles into $\tilde{S^0}$ and $\tilde{E}$ using weighting parameters $\lambda_g$ and $\lambda_m$. Then using reweighted $\tilde{E}$ to enhance reweighted $\tilde{S^0}$ into a disease-specific network $S_E$. At last, update initial disease-association gene and metabolite scores by diffusing on the enhanced and denoised multi-omics network $S_E$, and prioritize candidate disease-associated genes and metabolites simultanuously using updated disease-association gene and metabolite scores.

Installation

• The R package of MultiNEP can be installed through:
  if (!requireNamespace("devtools", quietly = TRUE))
install.packages("devtools")
library("devtools")
install_github("Karenxzr/MultiNEP")

Usage

It is quite simple to run MultiNEP through a wrapper function of nep. Input required:

• General network s0: an $n \times n$ matrix. With rownames and colnames being set as gene/metabolite names.
• Disease similarity matrix E: an $n \times n$ matrix. With rownames and colnames being set as gene/metabolite names. Note, all values in E should range from 0 - 1.
• Initial disease association scores signal: a dataframe with the first column being feature names, the second column being initial association scores. Input p-values as default association scores.
• Feature name list feature_name_list: a list with the first element containing all gene names and the second containing all metabolite names.

You can find sample input data within pacakge or in the data folder. See an application example as below:

library(MultiNEP)
results = nep(s0=s0,E=E,signals=signal,feature_name_list = feature_name_list, model='multinep')

Run results$vec to get prioritized candidate disease-associated multi-omics features. If you want to get re-weighted and enhanced disease-specific multi-omics network $S_E$, run results$enhanced_mat$unprocessed or results$enhanced_mat$processed with return_mat argument set as TRUE.

You can also change parameters such as $\lambda_g$ or $\lambda_m$. Run ?nep to find more details.

Reference

• Zhuoran Xu, Luigi Marchionni, Shuang Wang, MultiNEP: a multi-omics network enhancement framework for prioritizing disease genes and metabolites simultaneously, Bioinformatics, Volume 39, Issue 6, June 2023, btad333, https://doi.org/10.1093/bioinformatics/btad333