Teaching

Machine learning applied to biological data. The course opens with supervised and unsupervised learning, then works through the problems that make omics data awkward: high dimensionality, class imbalance, and integration across data levels. Students cover dimensionality reduction, gene selection from multi-level omics, random forests for imbalanced genomic classification, and optimisation methods for microarray feature selection. Later modules move into image segmentation, deep learning for medical imaging and diagnostics, and genome signal processing for hotspot identification.

Hands-on sequence and structure analysis. Students retrieve and manage data from the major biological databases, then work through pairwise alignment by global, local, and dot plot methods before moving to BLAST searches and multiple sequence alignment with phylogeny in Clustal Omega. The structural half covers motif and domain searching, the PDB, protein secondary structure prediction, and 3D structure visualisation.