How labioinfoufsc/ebscore works
Project type and purpose
• This is a data-science/ML research pipeline, not a web or microservices system.
• Purpose: ingest epitope data from IEDB/UniProt, preprocess it (via Java and EpiBuilder), train and evaluate a decision-tree model in a Jupyter notebook, and validate Trypanosoma epitopes.
File structure
• Top-level files: README.md (project description), epitryp.ipynb (analysis and modeling notebook), LICENSE, .gitignore.
• assets/ folder contains:
• RandomPeptide.java (Java code for generating random peptides)
• experiment.ows (Orange Data Mining workflow file)
• No “frontend” or “backend” folders—this is a single-pipeline project.
README for additional insights
• Data sources: IEDB and UniProtKB.
• Preprocessing tools: EpiBuilder and custom Java code in assets.
• Modeling and evaluation: Python libs (pandas, numpy, scikit-learn, seaborn, matplotlib) inside a Colab notebook; Orange workflow for algorithm comparison.
Essential architectural elements
• Keep components granular (separate Java preprocessing from Python modeling)
• External dependencies (databases, tools)
• Parallel or alternative paths (Orange vs. notebook evaluation)