Protein Subcellular Localisation Prediction Using Machine Learning – HPA MSc Project

This repository contains the datasets and scripts used in the analysis described in my dissertation on protein subcellular localization prediction. 📂 Contents 1️⃣ subcellular_location_filtered.tsv - The filtered Human Protein Atlas subcellular_location.tsv file used for model training. - Contains: Gene, Gene name, and subcellular location annotations (Enhanced, Supported, Approved, Uncertain). 2️⃣ cnn_model_notebook.ipynb - Jupyter notebook for training convolutional neural network (CNN) models. - Includes web scraping code for image URL retrieval, data processing steps, and model training routines. 3️⃣ transformer_model_notebook.ipynb - Jupyter notebook for training transformer-based models. - Includes web scraping code for image URL retrieval, data processing steps, and model training routines. 📌 Notes - The full image dataset (∼12 GB) used in this work is not included due to size limitations and lack of access to the original virtual machine where the dataset was stored. - The notebooks contain the code required to reproduce the image scraping and processing if access to the Human Protein Atlas is available. - No original image files are included in this repository in compliance with Human Protein Atlas data sharing policies. - Image URLs can be regenerated using the provided code and the Human Protein Atlas (https://www.proteinatlas.org/).