Prioritization of biological processes based on the reconstruction and analysis of associative gene networks describing the response of plants to adverse environmental factors

P S Demenkov,E A Oshchepkova,T V Ivanisenko,V A Ivanisenko

doi:10.18699/vj21.065

Abstract

Methods for prioritizing or ranking candidate genes according to their importance based on specif ic criteria via the analysis of gene networks are widely used in biomedicine to search for genes associated with diseases and to predict biomarkers, pharmacological targets and other clinically relevant molecules. These methods have also been used in other f ields, particularly in crop production. This is largely due to the development of technologies to solve problems in marker-oriented and genomic selection, which requires knowledge of the molecular genetic mechanisms underlying the formation of agriculturally valuable traits. A new direction for the study of molecular genetic mechanisms is the prioritization of biological processes based on the analysis of associative gene networks. Associative gene networks are heterogeneous networks whose vertices can depict both molecular genetic objects (genes, proteins, metabolites, etc.) and the higher-level factors (biological processes, diseases, external environmental factors, etc.) related to regulatory, physicochemical or associative interactions. Using a previously developed method, biological processes involved in plant responses to increased cadmium content, saline stress and drought conditions were prioritized according to their degree of connection with the gene networks in the SOLANUM TUBEROSUM knowledge base. The prioritization results indicate that fundamental processes, such as gene expression, post-translational modif ications, protein degradation, programmed cell death, photosynthesis, signal transmission and stress response play important roles in the common molecular genetic mechanisms for plant response to various adverse factors. On the other hand, a group of processes related to the development of seeds (“seeding development”) was revealed to be drought specif ic, while processes associated with ion transport (“ion transport”) were included in the list of responses specif ic to salt stress and processes associated with the metabolism of lipids were found to be involved specif ically in the response to cadmium.

Highlights

The rapid development of high-performance experimental methods has significantly expanded the ability to generate large sets of genomic, transcriptomic and proteomic data in scientific research
Using the SOLANUM TUBEROSUM knowledge base, we reconstructed the associative gene networks of A. thaliana that described the interaction of genes with biological processes related to plant responses to adverse environmental factors, including drought, salt stress and increased cadmium content
To study the potential molecular genetic mechanisms underlying the reconstructed gene networks, biological processes were prioritized based on the centrality of their interactions with the network’s genes/proteins

Summary

Introduction

The rapid development of high-performance experimental methods has significantly expanded the ability to generate large sets of genomic, transcriptomic and proteomic data in scientific research. Among the widely used approaches in computational analysis of gene sets identified in experiments are prioritization methods (Raj, Sreeja, 2018), which rank the studied genes (or other objects, such as di seases) by characterising their proximity to a set from a given learning sample. The higher the proximity in relative units, the greater the priority of the analysed object as a candidate possessing the same properties as objects in the training set. Such methods are used in biomedicine to detect candidate genes associated with diseases (Tranchevent et al, 2016), disease biomarkers (Jha et al, 2020), potential pharmacological targets (Cesur et al, 2020) and drug republic (Pushpakom et al, 2019). In animal husbandry and crop production, prioritization methods have been applied to analyse genomic data related to markeroriented and genomic selection (Arruda et al, 2016; Crossa et al, 2017; Kochetov et al, 2017; Kolchanov et al, 2017; Cai et al, 2019; Voss-Fels et al, 2019; Sun et al, 2020), as well as Raspanic loci analysis (Bargsten et al, 2014; Schaefer et al, 2018; Lin et al, 2019)

Methods

Results

Conclusion