Overview

Teaching: 60 min
Exercises: 120 min

Objectives

• Understand how biological information is used to predict hosts

• Understand the difficulties with host prediction

• Learn about the new techniques that are being used for host prediction

Host prediction lecture

We will start with a short lecture, then you can either read the abstract and introduction of the PhageTransformer preprint or the RaFAH paper. Both tools use a machine learning approach to solve the phage-host prediction problem. Please choose only one paper and afterwards we will discuss the differences.

Write a short summary

Please write half a page into your daily report about phage-host interactions and how we can predict them. You dont have to explain everything, you can focus on parts of what we went through. You can work along the following points:

1. What are some biological interactions viruses have with their hosts?
2. How can we exploit these interactions?
3. There are many methods for predicting phage-host interactions. What are some of the issues they run into?
4. How can you be confident in your host prediction?

Additional resources

• IBM link on Random Forest (RF): introduces RF and decision trees in a short and simple way
• Chapter on RF for Bioinformatics: explains how RF measures feature importance and describes Bioinformatic applications
• Chapter on Decision Trees and RF: describes decision trees and RF by direct application in python. Also, if you want a great introduction to machine learning in general and python programming, take a look at the whole book by Jake VanderPlas, which is freely available online

Key Points