Page 71 - GPD-1-2

Gene & Protein in Disease Recent advances and challenges of network biology

A

B

Figure 2. Biological network construction. (A) Approaches to construct biological networks. (B) Flowchart of data-driven biological network construction
and applications. Based on biological data, various mathematical and statistical models can be developed to realize network construction; the network
construction method should be further verified by datasets with known network structures, and subsequently, real-world networks can be constructed for
various applications based on experimentally detected datasets.

Table 1. Advantages and disadvantages of the four network construction approaches.
Approach Advantages Disadvantages
Online databases Easily obtained. Only applicable for limited organisms; some databases are not timely
updated.
Artificial algorithms Easily generated; evolutions of network Only limited to theoretical investigations; there are still gaps in real-world
features can be explored. biological networks.
Dynamical network theory Based on the theory of dynamical systems; Requires node dynamics and dynamical system theories; difficult to be
potentially applicable for real-world control. used in biological networks.
Data-driven approaches Various mathematical and statistical models Difficult to determine the correctness for organisms without any known
can be developed; explainable by data. network information; discrepancy among different methods; difficult to
determine the cutoff threshold values for certain methods.

bioinformatics [44,52,107,112,113] (Figure 3A). The guilt-by- 3.2.1. Identifying informative genes/proteins based
association principle assumes that genes/proteins that highly on biological network analysis
connected with disease genes tend to be disease ones [44,112,113] , Important node identification is a foundational topic in
whereas a basic assumption of the guilt-by-rewiring principle complex networks [4,56-61,114,115] . Nodes in complex networks
is that genes/proteins that altered their co-expression/ are heterogeneous, and different nodes generally play
interaction relationships under treatment are closely related different roles; it is significant and interesting to identify
to the causal phenotypes [52,107] . The two principles have been important nodes [114] . Depending on the types of networks
widely applied to identify important genes/proteins, predict and the concerned questions, important nodes have
the functions of genes/proteins, and explore the molecular different definitions. In biological contexts, important
mechanisms behind certain phenotypes. nodes largely represent informative genes/proteins in the

Volume 1 Issue 2 (2022) 5 https://doi.org/10.36922/gpd.v1i2.101