Genome Evolution: A computational approach
Spring 2008
Amos Tanay, Tuesdays,
A new course that will introduce you to genomes and their evolution with an emphasis on modern computational methods for inference in probabilistic models. The course main rationale is that with the advent of genomic technologies, the focus of computational molecular biology turns from phylogeny and analysis of proteins to complex modeling of evolution in heterogeneous genomic regions. Understanding evolution in such regions involves probabilistic models that are rich in parameters and structure and greatly extend over traditional methods.
Lecture 1: Modern challenges in evolution, Markov processes Lecture 2: Continuous time Markov processes, simple tree models: inference and learning Lecture 3: EM. More on inference. HMMs. Lecture 4: Beyond trees: sampling Lecture 5: Variational inference Lecture 6: Belief propagation Lecture 7: The Human genome; Brief evolutionary history of everything Lecture 8: Intro to population genetics Lecture 9: Mutations/Selection Lecture 10: Selection: proteins Lecture 11: Selection: binding sites Lecture 12: Selection: binding sites II Lecture 13: Selection: RNA/networks |
see 8 see 11 |
Notes:
2/6 Exam example is here
22/5 Exam topics + partial proofs are here
22/5 Ex4 updated
25/3 Yedida, Freeman and Weiss on GLBP here
20/3 Deadline for Ex2 is March 30
20/3 Clarifications for ex2 q2: The Markov models (1-order or 2-order) are defined by their transition probabilities. You are asked about sequences that were generated by sampling from the models for a long time, so you should assume you reached the stationary distribution, and there is no need in additional parameters – you can use P(x) (the stationary distribution) in yours answer without expressing it in terms of P(x->y).
18/3 correction to ex2 q2 posted – you should look for the minimum ratio not the maximum
18/3 An early tutorial on variational inference is here
Inference in phylo-hmm – the paper by Jojic, Geiger, Siepel, Haussler, Heckerman here
A free textbook that cover some inference techniques is here
11/3 Ex2 is on-line, due March 25
11/3 Suggested reading:
http://www.nature.com/nature/journal/v451/n7182/full/nature06633.html
11/3 Note: Ex1 due 16/3
9/3 NOTE: Ex1 due 13/3
-Typos in Ex1, Q6 corrected.
-In Q6 – it is enough to write down the correct EM formula.
- It is ok If you cannot solve the maximization problem with a closed form!
- Assume mutation probabilities are low (i.e. 2*Pr(x->y) is still small
3/3 Note that ex1 is due on March 11
3/3 Lecture 3 perlim slides are on-line