Network analysis of hepatocellular carcinoma liquid biopsies augmented by single-cell sequencing data

Abstract

Liquid biopsy, the analysis of body fluids, represents a promising approach for disease diagnosis and prognosis with minimal intervention. Sequencing cell-free RNA derived from liquid biopsies has been very promising for the diagnosis of several diseases. Cancer research, in particular, has emerged as a prominent candidate since early diagnosis has been shown to be a critical determinant of disease prognosis. Although high-throughput analysis of liquid biopsies has uncovered many differentially expressed genes in the context of cancer, the functional connection between these genes is not investigated in depth. An important approach to remedy this issue is the construction of gene networks which describes the correlation patterns between different genes, thereby allowing to infer their functional organization. In this study, we aimed at characterizing extracellular transcriptome gene networks of hepatocellular carcinoma patients compared to healthy controls. Our analysis revealed a number of genes previously associated with hepatocellular carcinoma and uncovered their association network in the blood. Our study thus demonstrates the feasibility of performing gene co-expression network analysis from cell-free RNA data and its utility in studying hepatocellular carcinoma. Furthermore, we augmented cell-free RNA network analysis with single-cell RNA sequencing data which enables the contextualization of the identified network modules with cell-type specific transcriptomes from the liver.

Deep Learning-Based, Multiclass Approach to Cancer Classification on Liquid Biopsy Data

“…There are many breakthrough cancer studies related to analyzing those biomarkers where each type may shine in its field, covering the following areas: a) screening and early cancer detection research (Epi-proColon, GRAIL) [6], [7]; b) tools for decision-making and companion diagnostics (Cobas EGFR Mutation Test v2, Guardant360) [8], [9]; as well as c) determination whether the patient has a drug targetable mutation (FoundationOne-Liquid, Resolution-HRD) [10]. Multi-omics models and multiplatform solutions have also emerged, basing their outcomes on data from multiple related sources [11], [12], [13], [14]. In general, liquid biopsies constitute a promising alternative to traditionally performed JTEHM-00138-2023 tissue biopsies and, in certain conditions, can complement, if not replace, traditional biopsy evaluation [15], [16].…”