Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Integrating Protein Sequence and Expression Level to Analysis Molecular Characterization of Breast Cancer Subtypes

Hossein Sholehrasa, Majid Jaberi-Douraki

TL;DR

This study tackles the challenge of breast cancer heterogeneity by integrating protein sequence embeddings with protein expression data to better characterize molecular subtypes and predict clinical outcomes. Using ProtGPT2, protein sequences are embedded into high-dimensional representations that are modulated by protein expression levels and reduced via PCA, then combined with clinical features for clustering and outcome prediction through ensemble K-means and XGBoost. The approach yields biologically meaningful patient clusters and high predictive performance for survival (F1 = 0.88) and biomarker status (F1 = 0.87), with key proteins such as KMT2C, CLASP2, and MYO1B highlighted as influential in hormone signaling and therapy resistance. Protein–protein interaction networks further reveal interconnected modules that may drive subtype-specific tumor behavior, underscoring the potential of sequence-plus-expression embeddings to enhance precision oncology for breast cancer and other complex diseases.

Abstract

Breast cancer's complexity and variability pose significant challenges in understanding its progression and guiding effective treatment. This study aims to integrate protein sequence data with expression levels to improve the molecular characterization of breast cancer subtypes and predict clinical outcomes. Using ProtGPT2, a language model specifically designed for protein sequences, we generated embeddings that capture the functional and structural properties of proteins. These embeddings were integrated with protein expression levels to form enriched biological representations, which were analyzed using machine learning methods, such as ensemble K-means for clustering and XGBoost for classification. Our approach enabled the successful clustering of patients into biologically distinct groups and accurately predicted clinical outcomes such as survival and biomarker status, achieving high performance metrics, notably an F1 score of 0.88 for survival and 0.87 for biomarker status prediction. Feature importance analysis identified KMT2C, CLASP2, and MYO1B as key proteins involved in hormone signaling, cytoskeletal remodeling, and therapy resistance in hormone receptor-positive and triple-negative breast cancer, with potential influence on breast cancer subtype behavior and progression. Furthermore, protein-protein interaction networks and correlation analyses revealed functional interdependencies among proteins that may influence the behavior and progression of breast cancer subtypes. These findings suggest that integrating protein sequence and expression data provides valuable insights into tumor biology and has significant potential to enhance personalized treatment strategies in breast cancer care.

Integrating Protein Sequence and Expression Level to Analysis Molecular Characterization of Breast Cancer Subtypes

TL;DR

This study tackles the challenge of breast cancer heterogeneity by integrating protein sequence embeddings with protein expression data to better characterize molecular subtypes and predict clinical outcomes. Using ProtGPT2, protein sequences are embedded into high-dimensional representations that are modulated by protein expression levels and reduced via PCA, then combined with clinical features for clustering and outcome prediction through ensemble K-means and XGBoost. The approach yields biologically meaningful patient clusters and high predictive performance for survival (F1 = 0.88) and biomarker status (F1 = 0.87), with key proteins such as KMT2C, CLASP2, and MYO1B highlighted as influential in hormone signaling and therapy resistance. Protein–protein interaction networks further reveal interconnected modules that may drive subtype-specific tumor behavior, underscoring the potential of sequence-plus-expression embeddings to enhance precision oncology for breast cancer and other complex diseases.

Abstract

Breast cancer's complexity and variability pose significant challenges in understanding its progression and guiding effective treatment. This study aims to integrate protein sequence data with expression levels to improve the molecular characterization of breast cancer subtypes and predict clinical outcomes. Using ProtGPT2, a language model specifically designed for protein sequences, we generated embeddings that capture the functional and structural properties of proteins. These embeddings were integrated with protein expression levels to form enriched biological representations, which were analyzed using machine learning methods, such as ensemble K-means for clustering and XGBoost for classification. Our approach enabled the successful clustering of patients into biologically distinct groups and accurately predicted clinical outcomes such as survival and biomarker status, achieving high performance metrics, notably an F1 score of 0.88 for survival and 0.87 for biomarker status prediction. Feature importance analysis identified KMT2C, CLASP2, and MYO1B as key proteins involved in hormone signaling, cytoskeletal remodeling, and therapy resistance in hormone receptor-positive and triple-negative breast cancer, with potential influence on breast cancer subtype behavior and progression. Furthermore, protein-protein interaction networks and correlation analyses revealed functional interdependencies among proteins that may influence the behavior and progression of breast cancer subtypes. These findings suggest that integrating protein sequence and expression data provides valuable insights into tumor biology and has significant potential to enhance personalized treatment strategies in breast cancer care.
Paper Structure (13 sections, 10 equations, 4 figures, 2 tables)

This paper contains 13 sections, 10 equations, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Data Source
  4. Preprocessing
  5. Integrating Protein Expression level with Protein Sequence
  6. Clustering and Network Analysis
  7. Classification
  8. Results
  9. Classifications
  10. Clustering
  11. Discussion
  12. Conclusion
  13. Supplementary Figures

Figures (4)

  • Figure 1: Pearson correlation matrix of the top 10 important proteins in the biomarkers prediction
  • Figure 2: Protein-protein interactions based on the STRING-db of the top 10 important proteins
  • Figure 3: Dendrogram resulting from the ensemble clustering of the patient dataset combined with hierarchical clustering for each patient case
  • Figure A.1: Co-occurrence matrices for three patient clusters (Orange, Green, and Red), illustrating correlations between breast cancer markers.