The goal of this project is to attempt to predict the thermal stability of various enzymes given their amino acid sequence. The task was part of a competition hosted by Novozymes on Kaggle: https://www.kaggle.com/competitions/novozymes-enzyme-stability-prediction/overview
We intend to use the transformers framework from Huggingface.
The dataset is taken from https://www.kaggle.com/competitions/novozymes-enzyme-stability-prediction/data and consists of roughly 31 thousand samples. The samples consist of a unique sequence id, an amino acid sequence, a pH value that the amino acid was treated under, a data source (not used) and finally the prediction target, which is an unspecified continuous value where high values indicate a high thermal stability and vice versa.
The pretrained model we apply is a BERT model trained on amino acid sequences. The model was trained to finish a sequence of amino acids given the initial sequence. This task requires predicting a continuous value, so the BERT model is extended with a multi-layer perceptron.
Currently inference with the BERT model is computationally expensive, so unless we find a way to optimize this we will move on with another model
We intend to use the transformers framework from Huggingface.
The dataset is taken from https://www.kaggle.com/competitions/novozymes-enzyme-stability-prediction/data and consist of roughly 31 thousand samples. The samples consist of a unique sequence id, an amino acid sequence, a pH value that the amino acid was treated under, a data source (not used) and finally the prediction target, which is an unspecified continuous value where high values indicate a high thermal stability and vice versa.
#To retrieve the data from google_storage (mediated through dvc) please just:
git clone https://github.com/AlexanderVoldby/02476_enzyme_stability.git -b dev_pau
cd 02476_enzyme_stability
dvc pull
The pretrained model we apply is a BERT model trained on amino acid sequences. The model was trained to finish a sequence of amino acids given the initial sequence. This task requires predicting a continuous value, so the BERT model is extended with a regression layer.
The directory structure of the project looks like this:
├── Makefile <- Makefile with convenience commands like `make data` or `make train`
├── README.md <- The top-level README for developers using this project.
├── data
│ ├── processed <- The final, canonical data sets for modeling.
│ └── raw <- The original, immutable data dump.
│
├── docs <- Documentation folder
│ │
│ ├── index.md <- Homepage for your documentation
│ │
│ ├── mkdocs.yml <- Configuration file for mkdocs
│ │
│ └── source/ <- Source directory for documentation files
│
├── models <- Trained and serialized models, model predictions, or model summaries
│
├── notebooks <- Jupyter notebooks.
│
├── pyproject.toml <- Project configuration file
│
├── reports <- Generated analysis as HTML, PDF, LaTeX, etc.
│ └── figures <- Generated graphics and figures to be used in reporting
│
├── requirements.txt <- The requirements file for reproducing the analysis environment
|
├── requirements_dev.txt <- The requirements file for reproducing the analysis environment
│
├── tests <- Test files
│
├── mlops_enzyme_stability <- Source code for use in this project.
│ │
│ ├── __init__.py <- Makes folder a Python module
│ │
│ ├── data <- Scripts to download or generate data
│ │ ├── __init__.py
│ │ └── make_dataset.py
│ │
│ ├── models <- model implementations, training script and prediction script
│ │ ├── __init__.py
│ │ ├── model.py
│ │
│ ├── visualization <- Scripts to create exploratory and results oriented visualizations
│ │ ├── __init__.py
│ │ └── visualize.py
│ ├── train_model.py <- script for training the model
│ └── predict_model.py <- script for predicting from a model
│
└── LICENSE <- Open-source license if one is chosenCreated using mlops_template, a cookiecutter template for getting started with Machine Learning Operations (MLOps).
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent