Sarcoma-TC-Clusters

This repository contains scripts and data for the analysis for the publication "Machine learning-based analysis of genomic and transcriptomic data unveils sarcoma clusters with superlative prognostic and predictive value".

---

Introduction

Sarcomas are a heterogeneous group of cancers with distinct genetic profiles. This repository explores the identification and validation of transcriptomic clusters in sarcoma.

Note - Proprietary dataset from IPO used can be provided upon request

The repository leverages TCGA-SARC data and integrates external validation datasets to ensure reproducibility and robustness of the results.

---

Folder Structure

├── FILES
│   ├── Proprietary files can be sent upon request
├── RESULTS
│   ├── TCGA-TC-Classified.csv # TCGA-SARC Patients Classified with TC
│   ├── Consensus_Clusters_Genes.txt # Genes discovered from Proprietary Cohort that define clusters
│   ├── clinical_cinsarc_paper.csv # TCGA-SARC classified with CINSARC
│   ├── Sarculator_TCGA.csv # TCGA-SARC Classified with SARCULATOR
│   ├── TCGA-SARC.csv # TCGA-SARC Gene Expression Matrix
│   └── Sarculator_Results.xlsx # Proprietary patients classified with TC
├── *.R                               # Main R scripts for analysis
├── .gitignore                        # Ignore configuration
└── README.md                         # Project documentation

---

Dependencies

To install dependencies please run 0.install_packages.R script

This code was run in RStudio (Version 2024.04.1+748 (2024.04.1+748)

---

Usage

Scripts must be run in ascendant order

0. Data Preprocessing

The script 0.clinical_data_preprocessing.R processes and cleans the clinical data and RNA expression data.

1. Consensus Clustering

Run 1.consensus_cluster.R to identify transcriptomic clusters and extract cluster-specific genes.

2. Survival Analysis

Use 2.survival_analysis.R to evaluate the clinical relevance of the identified clusters using Kaplan-Meier survival plots and Cox proportional hazard models.

3. TCGA-SARC Validation

Scripts in the 3.*.R series validate the clusters using TCGA-SARC, integrating and comparing with nomogram tools - Sarculator and gene expression signatures - CINSARC.

Citation

Esperança-Martins, M., Vasques, H., Ravasqueira, M. S., Lemos, M. M., Fonseca, F., Coutinho, D., López, J. A., Huang, R. S. P., Dias, S., Gallego-Paez, L., Costa, L., Abecasis, N., Gonçalves, E., & Fernandes, I. (2025). Machine learning-based analysis of genomic and transcriptomic data unveils sarcoma clusters with superlative prognostic and predictive value. medRxiv. https://doi.org/10.1101/2025.01.31.25321492

---

License

This repository is licensed under the MIT License. See LICENSE for details.