Chapter 10 of Bioinformatics

Machine learning on sequences soon

From k-mers and one-hot encodings to convolutional networks and protein language model embeddings, with sequence-identity splits as the honest evaluation.

soon This chapter is planned and not yet written. The outline below is the current plan; it may change.

What it will cover

• k-mer counting and one-hot encoding of sequences
• Sequence-identity splits and why random splits lie
• 1-D convolutional networks for motifs
• Protein language model embeddings (ESM-style) as features
• Function and localisation prediction tasks
• Lab 10: embed a protein family and predict a property with identity-aware CV

Planned learning outcomes

1. Turn sequences into feature matrices in several ways and compare them.
2. Train a small CNN on sequence data and visualise what it learned.
3. Use pre-trained protein embeddings with an honest sequence-identity split.

In the meantime

Everything here builds on Part I, in particular the evaluation habits of Chapter 2. Episode 2.6 already introduces the group splits that become sequence-identity splits here.

PreviousChapter 9 Biological sequences and databasesNextChapter 11 Omics data

Edit this pageReport a problemLast updated 2026-11-02