Welcome to genomeNLP’s documentation!

Copyright (c) 2022 Tyrone Chen , Navya Tyagi , Sarthak Chauhan, Anton Y. Peleg , and Sonika Tyagi .

Code in this repository is provided under a MIT license. This documentation is provided under a CC-BY-3.0 AU license.

Visit our lab website here. Contact Sonika Tyagi at sonika.tyagi@monash.edu.

Note

The main repository is on github but also mirrored on gitlab. Please submit any issues to the main github repository only.

Contents:

• genomeNLP: Genome recoding for Machine Learning Usage incorporating genomicBERT
  • Highlights
  • Cite us with:
  • Install
    • Mamba (automated)
    • Mamba (manual)
  • Usage
    • 0. Available commands
    • 1. Preprocessing
    • 2. Classification
    • 3. Comparing deep learning models trained by genomicBERT
    • 4. Case study
  • Background
  • Acknowledgements
• genomeNLP: Case study of deep learning
  • Outline
    • Learning objectives
    • Potential/preferred prerequisite knowledge
    • Glossary
  • 1. Introduction
    • What is NLP and genomics
    • Why apply NLP in genomics
    • Distinction between conventional NLP and genome NLP
  • 2. Connect to a remote server
  • 3. Installing conda, mamba and genomenlp
  • Case studies per molecule type
    • DNA case study
    • RNA case study
    • Protein case study
  • Citation
• genomeNLP: Case study of DNA
  • 4. Setting up a biological dataset
  • 5. Format a dataset for input into genomeNLP
  • 6. Preparing a hyperparameter sweep
  • 7. Selecting optimal hyperparameters for training
  • 8. With the selected hyperparameters, train the full dataset
  • 9. Perform cross-validation
  • 10. Compare different models
  • 11. Obtain model interpretability scores
  • Citation
• genomeNLP: Case study of Protein
  • 4. Setting up a biological dataset
  • 5. Format a dataset for input into genomeNLP
  • 6. Preparing a hyperparameter sweep
  • 7. Selecting optimal hyperparameters for training
  • 8. With the selected hyperparameters, train the full dataset
  • 9. Perform cross-validation
  • 10. Compare different models
  • 11. Obtain model interpretability scores
  • Citation
• Create a token set from sequences
  • Source data
  • Results
    • Empirical tokenisation
    • Conventional k-mers
  • Notes
  • Usage
    • Empirical tokenisation
    • Conventional k-mers
• Create a dataset object from sequences
  • Source data
  • Results
  • Notes
  • Usage
• Create embeddings from a tokenised dataset
  • Source data
  • Results
    • Empirical tokenisation
    • Conventional k-mers
  • Notes
  • Usage
    • Empirical tokenisation
    • Conventional k-mers
• Perform a hyperparameter sweep
  • Source data
  • Results
    • Deep learning
    • Frequency-based approaches
    • Embedding
  • Notes
  • Usage
    • genomicBERT: Deep learning
    • Frequency based approach
    • Embedding based approach
• genomicBERT: Train a deep learning classifier
  • Source data
  • Results
  • Notes
  • Usage
• Perform cross-validation
  • Source data
  • Results
    • Deep learning
    • Frequency-based approaches
    • Embedding
  • Notes
  • Usage
    • Deep learning
• Compare performance of different deep learning models
  • Source data
  • Results
  • Notes
  • Usage
• Generate synthetic sequences for use in classification
  • Source data
  • Results
  • Notes
  • Usage
• Get class attribution for deep learning models
  • Source data
  • Results
    • Deep learning
  • Notes
  • Usage
    • genomicBERT: Deep learning

Indices and tables

• Index

• Module Index

• Search Page