Course BIO120
Machine Learning Techniques for Bioinformatics
Machine Learning Techniques for Bioinformatics
Duration: 5 Days
Intended Audience
Attendees are expected to have a good working knowledge of statistics, calculus and algebra as might be gained from completing a science degree with a substantial mathematical component.
Course Overview
After a brief overview of the underlying molecular biology concepts, and of basic probability theory and frameworks for probabilisting reasoning and modeling the course goes on to survey the various machine learning algorithms that have been developed over the last 50 years. Specific algorithms and techniques such as Neural Networks, Hiddent Markov Models, Probabilistic Graphical Models and Stochastic grammars are then explored in greater detail The course is predominantly a taught course. For those who are willing to put in the extra hours there will be opportunities to do some lab and project work.
Course Benefits
After completing this course attendees should have a good knowledge of machine learning techniques as they apply to Bioinformatics.
Course Contents
Foundations - Molecular Biology
- Digital symbol sequences for representing biological data
- Overview of genomes, proteins and proteomes
- Information content of biological sequences
Foundations - Machine Learning
- Basics of Bayesian modeling
- Bayesian inference and induction
- Probabilistic modeling and inference
Machine learning algorithms - an overview
- Dynamic programming
- Gradient descent
- Simulated annealing
- EM (Expectation Maximisation) and GEM (Generalised Expectation Maximisation) algorithms
- Markov-Chain and Monte-Carlo methods
- Evolutionary and Genetic Algorithms
- Learning algorithms
- Neural networks
Applications of Neural Networks to Bioinformatics
- Protein secondary structure prediction
- Prediction of signal peptides and their cleavage sites
- Eukaryotic gene finding and Intron Splice Site prediction
Applications of Hidden Markov Models (HMMs) to Bioinformatics
- Learning algorithms based on HMMs
- Applications of HMMs to proteins - structure and function
- Applications of HMMs to DNA and RNA - structure and function
Applications of Probabilistic graphical models in Bioinformatics
- Markov Models and DNA symmetries
- Markov Models and gene location
- Bidirectional recurrent neural networks for protein secondary structure prediction