Identify Statistical Similarities and Differences Between the Deadliest Cancer Types Through Gene Expression

TL;DR

This study compares gene expression networks across five deadly cancer types to identify shared and distinct molecular features, aiding in better classification and treatment strategies.

Contribution

It introduces a comparative network analysis approach using WGCNA and correlation networks across multiple cancer types, revealing key similarities and differences.

Findings

Identified critical genes differentiating cancer types

Revealed shared network properties among cancers

Highlighted unique network features for each cancer

Abstract

Prognostic genes have been well studied within each type of cancer. However, investigations of the similarities and differences across cancer types are rare. In view of the optimal course of treatment, the classification of cancers into subtypes is critical to the diagnosis. We examined the properties in gene co-expression networks using a patient-to-patient correlation network analysis and a weighted gene correlation network analysis (WGCNA) for five cancer types using data generated by UC Irvine. We further analyze and compare the degree, centrality and betweenness of the network for each cancer type and apply a multinomial logistic regression to identify the critical subset of genes. Given the cancer types provided, our study presents a view of emergent similarities and differences across cancer types.