Breast cancer histopathology image-based gene expression prediction using spatial transcriptomics data and deep learning.

Tumour heterogeneity in breast cancer poses challenges in predicting outcome and response to therapy. Spatial transcriptomics technologies may address these challenges, as they provide a wealth of information about gene expression at the cell level, but they are expensive, hindering their use in large-scale clinical oncology studies. Predicting gene expression from hematoxylin and eosin stained histology images provides a more affordable alternative for such studies. In this repository, we present the code implementation of BrST-Net, a deep learning framework for predicting gene expression from histopathology images using spatial transcriptomics data. The methodology outperforms previous studies, achieving positive correlations for a larger number of genes and higher correlation coefficients.

Contents: This repository contains the following Python files:

01_file_organizer.py: Script for organizing the input data files.

02_stain_normalization.py: Script for stain normalization of histology images.

03_spatial_gene_analysis.py: Script for spatial gene analysis using transcriptomics data.

04_generating_train_test.py: Script for generating train and test datasets.

get_patches_with_different_resolution.py: Script for obtaining patches with different resolutions.

enselbl_identity.py: Script for working with Ensembl IDs.

Br_STNet_baseline.py: Main script for the BrST-Net deep learning framework for gene expression prediction.

Please refer to the paper for a detailed explanation of the methodology and the results achieved.