Bookmark and Share

Algorithms in bioinformatics, PhD


Technical University of Denmark


General course objectives:
To provide the student with an overview and in-depth understanding of bioinformatics machine-learning algorithms. Enable the student to first evaluate which algorithm(s) are best suited for answering a given biological question and next implement and develop prediction tools based on such algorithms to describe complex biological problems such as immune system reactions, vaccine discovery, disease gene finding, protein structure and function, post-translational modifications etc.

Learning objectives:
A student who has met the objectives of the course will be able to:
  • understand the details of the algorithms commonly used in bioinformatics.
  • develop computer programs implementing these algorithms.
  • identify which type of algorithm is best suited to describe a given biological problem.
  • understand the concepts of data redundancy and homology reduction.
  • develop bioinformatics prediction algorithms describing a given biological problem.
  • implement and develop prediction tools, on a detailed level, using the following algorithms: Dynamic programming, Sequence clustering, Weight matrices, Artificial neural networks, and Hidden Markov models.
  • design a project where a biological problem is analyzed using one or more machine learning algorithms.
  • implement, document and present the course project.

Contents:
The course will cover the most commonly used algorithms in bioinformatics. Emphasis will be on the precise mathematical implementation of the algorithms in terms of functional computer programs. During the course, biological problems will be introduced and analyzed with the purpose of highlighting the strengths and weaknesses of the different algorithms. The following topics will be covered: Dynamic programming: Needleman-Wunsch, Smith-Waterman, and alignment heuristics Data redundancy and homology reduction: Hobohm and other clustering algorithms Weight matrices: Sequence weighting, pseudo count correction for low counts, Gibbs sampling, and Psi-Blast Hidden Markov Models: Model construction, Viterbi decoding, posterior decoding, and Baum-Welsh HMM learning Artificial neural networks: Architectures and sequence encoding, feed-forward algorithm, back propagation and deep neural networks The course will consist of lectures, discussion sessions and computer exercises, where the students will be introduced to the different algorithms, their implementation and use in analyzing biological problems. In the end of the course, the student will work on a group project were one or more of the algorithms introduced in the course are applied to analyze a biological problem of interest. The project report shall be written as a research paper including an in-depth review of the field covered by the project.

Back
Course organizer
Morten , Carolina Mercedes
Place/Venue
Anker Engelunds Vej 1
City
2800 Kgs. Lyngby
Country
Denmark
Workload
5
Link
http://kurser.dtu.dk/course/22175