https://github.com/elcaiseri/novozymes-enzyme-stability-prediction

21st Place Solution - Representation Is All You Need
https://github.com/elcaiseri/novozymes-enzyme-stability-prediction

chemistry machine-learning novozymes xgboost

Last synced: 7 days ago
JSON representation

21st Place Solution - Representation Is All You Need

• Host: GitHub
• URL: https://github.com/elcaiseri/novozymes-enzyme-stability-prediction
• Owner: elcaiseri
• Created: 2023-03-14T15:12:24.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2023-07-27T16:08:58.000Z (over 1 year ago)
• Last Synced: 2024-01-28T09:42:52.905Z (10 months ago)
• Topics: chemistry, machine-learning, novozymes, xgboost
• Language: Jupyter Notebook
• Homepage: https://www.kaggle.com/competitions/novozymes-enzyme-stability-prediction/discussion/393308
• Size: 172 KB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Novozymes Enzyme Stability Prediction

![Novozymes](https://img.shields.io/badge/Novozymes-blue.svg)

![Kaggle](https://img.shields.io/badge/Kaggle-blue.svg)

![XGBoost](https://img.shields.io/badge/XGBoost-blue.svg)

![PyTorch](https://img.shields.io/badge/PyTorch-blue.svg)

This repository contains the 21st place solution for the Novozymes Enzyme Stability Prediction competition. The goal of the competition is to help identify thermostable mutations in enzymes.

## Table of Contents

- [About](#about)

- [Codeflow](#codeflow)

- [Workspace](#workspace)

- [Dataset](#dataset)

- [Kaggle Link](#kaggle-link)

## About

This repository contains the 21st place solution for the Novozymes Enzyme Stability Prediction competition. The solution is based on the idea "**Representation Is All You Need.**"

![final-stage](./assets/final-stage.jpg)

For more information about the competition, idea, experiments, and results, see the [discussion here](https://www.kaggle.com/competitions/novozymes-enzyme-stability-prediction/discussion/393308).

## Codeflow

The solution is divided into three stages:

1. ThermoNet predictions.

2. Extract representations and XGBoost predictions.

3. Ensemble with different features.

## Workspace

The code and solution were developed in a Kaggle notebook.

## Dataset

The dataset used in this competition is from the Kaggle competition itself.

## Kaggle Link

Find more about the solution and the code at [Kaggle@elcaiseri](https://www.kaggle.com/elcaiseri).