JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

OVsignGenes: A Gene Expression-Based Neural Network Model Estimated Molecular Subtype of High-Grade Serous Ovarian Carcinoma

  • 0Engelhardt Institute of Molecular Biology, Russian Academy of Sciences, 119991 Moscow, Russia.
Cancers +

|

December 17, 2024

|

December 17, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

High-grade serous carcinomas (HGSCs) exhibit molecular heterogeneity. A new deep neural network model, OVsignGenes, accurately classifies HGSC subtypes, identifying the differentiated subtype as a potential tumor initiator.

Area Of Science

  • Oncology
  • Genomics
  • Bioinformatics

Background

  • High-grade serous carcinomas (HGSCs) are molecularly heterogeneous, impacting patient outcomes and treatment response.
  • Four previously defined subtypes (differentiated, immunoreactive, mesenchymal, proliferative) contribute to this complexity.
  • Accurate subtyping is crucial for personalized medicine and targeted therapy development in HGSCs.

Purpose Of The Study

  • To develop a robust method for classifying HGSC molecular subtypes.
  • To identify key differentially expressed genes associated with each subtype.
  • To investigate the potential role of the differentiated subtype in tumor progression.

Main Methods

  • Analysis of bulk RNA-seq, scRNA-seq, and spatial transcriptomic data from six cohorts (535 samples).
  • Differential gene expression analysis using edgeR and pathway enrichment analysis (KEGG, GSVA).
  • Development of a deep neural network (OVsignGenes) using Keras and TensorFlow for subtype prediction.

Main Results

  • Identification of 357 subtype-specific differentially expressed genes (96 differentiated, 33 immunoreactive, 91 mesenchymal, 137 proliferative).
  • The OVsignGenes model demonstrated high accuracy (AUC = 0.969) and platform resistance.
  • Successful validation of the model across multiple cohorts, including single-cell and spatial transcriptomic data.

Conclusions

  • The differentiated subtype, situated at the intersection of other subtypes, lacks unique gene expression and pathway profiles.
  • The differentiated subtype is proposed as an initiating subtype that evolves into other HGSC subtypes.
  • This finding has implications for understanding HGSC heterogeneity and developing targeted therapeutic strategies.

Keywords: