Introduction to Machine Learning for the Life Sciences (R version)

This repository contains the R code and materials for the Machine learning in the life sciences workshop, presented by the Australian BioCommons and led by Dr. Benjamin Goudey.

The goal of this workshop is to provide a high-level, hands-on introduction to machine learning (ML). It covers what ML is, its advantages and disadvantages compared to traditional modelling, and scenarios where it is a useful tool. Using example datasets, we explore commonly used algorithms, their trade-offs, and common pitfalls in how ML is applied and evaluated, with a specific focus on life science applications.

Repository Contents

• IntroToML2024_R.ipynb: This is the primary Jupyter notebook containing all the R code for the workshop's hands-on activities. It is designed to be run either in Google Colab or on your local machine.
• Workshop Slides/References: A PDF copy of the workshop slides and links to other useful resources.

---

How to Run the Code

You can run the workshop notebook IntroToML2024_R.ipynb in two ways:

1. Google Colab (Recommended): This is the easiest way to get started, as it requires no local setup.
2. Locally: You can run the notebook on your own computer if you have an R environment set up.

1. Using Google Colab

This method requires a Google account. We strongly recommend you complete the setup steps before the workshop, as some steps can take up to 20 minutes.

A. Open the Notebook in Colab

1. Navigate to https://colab.research.google.com/.

2. Sign in to your Google account.

3. Go to File → Open notebook.

4. Select the 'GitHub' tab.

5. Paste this URL into the search box: https://github.com/bwgoudey/IntroMLforLifeScienceWorkshopR/tree/main.

6. Click the search icon, and then click on IntroToML2024_R.ipynb to open it.

7. (Optional but recommended) Save a copy to your own Google Drive by clicking 'Copy to Drive'. This will allow you to save any changes you make.

B. Set Up the Environment (Allow ~20 mins)

Once the notebook is open, you must run two code cells to install the necessary R packages.

1. Navigate to 'Section 0: Package loading and data set up'.

2. Run the first code block (starting with # Google Colab set up) by clicking the 'play' icon. This will take about 1 minute.

3. Once it's finished, skip the second block and run the third code block (starting with # Load packages).

4. This block can take up to 20 minutes to run.

5. Once both cells have a green tick, your environment is ready.

2. Running Locally

You can also run the IntroToML2024_R.ipynb notebook on your local machine. This requires you to have R and Jupyter (e.g., via RStudio, VS Code, or JupyterLab) installed.

This workshop assumes familiarity with R and the tidyverse collection of packages.

You will need to install the following R package dependencies, which are listed in the notebook's setup cell:

• tidyverse
• rpart.plot
• MASS
• scales
• remotes
• tidymodels
• DataExplorer
• pROC
• randomForest
• ranger
• glmnet
• devtools

When running locally, you will need to adapt the code cells in 'Section 0' to suit your local environment (i.e., do not run the Colab-specific setup code).

---

Acknowledgements

This workshop is presented by the Australian BioCommons.

Thank you to Melissa Burke, Vicky Perreau, Christina Hall, and Giorgia Mori for their valuable input into this workshop material.