Page 84 - EJMO-9-3

Eurasian Journal of
Medicine and Oncology Mapping breast cancer PPI networks for targets

Despite advances in genomic and proteomic technologies, proteins onto the human interactome and analyzing
significant gaps remain in understanding the structural and their interactions using graph theory and metric space
functional organization of these networks, particularly how principles. This enables the identification of key proteins
their topological properties influence disease progression and pathways that drive breast cancer progression, as
and therapeutic resistance. Current research often focuses well as the exploration of their potential as therapeutic
2
on individual proteins or pathways, ignoring the systemic targets. By integrating pathway enrichment analysis and
interplay within the network. This study addresses this gap functional annotation, we provide a comprehensive view
by employing graph theory and metric space modeling to of the network’s structure and its implications for cancer
unravel the hierarchical organization of the breast cancer biology. This study not only advances our understanding
protein-protein interaction network (BCPIN), providing of breast cancer PPIs but also highlights the potential
insights into its structural and functional properties. of metric space modeling as a tool for uncovering novel
Graph theory has emerged as a powerful tool for therapeutic opportunities.
analyzing PPIs, enabling the representation of proteins as 2. Materials and methods
nodes and their interactions as edges. This approach has
3
been successfully applied to various cancers, including 2.1. Generation of cancer graphs
breast cancer, lung cancer, and glioblastoma, revealing To construct a crossover network of cancer-associated
6
7
4,5
key nodes that drive tumor progression and influence proteins, our approach involves mapping tumor-expressed
patient survival. For instance, past studies have identified proteins onto the human interactome (Figure 1). The
critical proteins involved in immune evasion, drug process is outlined as follows:
resistance, and metastatic signaling in hepatocellular (i) Build a graph, G, based on the binary PPI data from
carcinoma. Furthermore, common key nodes across the human protein network
8,9
multiple cancer types have been proposed as potential (ii) Identify a list, L, of tumor-expressed proteins that are
targets for broad-spectrum therapies. 10-14 However, these consistently observed across cancers
studies often lack a systematic framework to quantify (iii) For proteins present in list L, extract the relevant
the spatial and functional relationships between proteins interaction data from graph G
within the network. This limitation underscores the need (iv) Generate the final cancer-specific graph, G’.
for advanced modeling techniques, such as metric space
analysis, to better understand the network’s organization 2.2. Analyzing BCPINs as metric spaces
and identify therapeutic targets. In analysis, the BCPINs were treated as metric spaces and
Metric space modeling offers a novel approach to the distance between nodes was examined using graph
studying PPIs by quantifying the distances between nodes theory. One way to accomplish this was using a Python
within the network. In this framework, the network is wrapper for the C++ Boost Graph Library (http://www.
treated as a metric space, where the distance between two boost.org/) and implementing the Dijkstra algorithm.
nodes is defined by the shortest path connecting them. This method calculated the shortest distances between
15
This method allows for the classification of proteins into protein pairs within the network. Proteins with the
distinct zones based on their proximity to the network’s smallest maximum distance to their neighboring nodes
topological center, revealing a hierarchical structure that were identified as the central nodes of the network. By
reflects functional specialization. Previous studies have employing this methodology, the nodes were categorized
demonstrated the advantage of metric space modeling and segmented according to proximity from the center.
in identifying central hubs and peripheral nodes in
biological networks, providing insights into their roles in 2.3. Sources of PPI data
disease mechanisms. 16,17 For instance, central nodes often We analyzed a human PPI network consisting of 9448
correspond to highly connected proteins that regulate nodes and 181706 connections. 21
critical pathways, while peripheral nodes may represent
specialized or context-dependent proteins. By applying 2.4. Gene expression datasets in cancer
18
this approach to BCPIN, we aim to uncover the spatial For our study on breast cancer, we utilized gene expression
organization of proteins and their functional relevance in data from the Gene Expression Barcode database. The
breast cancer. dataset was accessed from OBRC: Online Bioinformatics
In this study, we formally constructed and modeled Resources Collection (Health Sciences Library System). 1
BCPIN as a metric space, building on previous successes in To maintain consistency, we selected the “unified tissue”
applying this framework to other biological networks. 19,20 1 https://www.hsls.pitt.edu/obrc/index.php?page=URL2
Our approach involves mapping tumor-expressed 0110523150503#

Volume 9 Issue 3 (2025) 76 doi: 10.36922/ejmo.8208