Group:Student Seminar
Title: Computational Biology
Speaker: Nathaniel Hobbs University
Time: 2010-11-18 16:00-2010-11-18 17:00
Venue: FIT 1-222

Abstract:

It has been observed in biology that some proteins (and by extension, the genes that produce them) are functionally related, and that therefore some genes might not mutate independently of each other.  In these cases, if two genes are "connected" somehow, the mutation of one without the other should happen less frequently (rather, they should mutate together).  In the talk, I will discuss how this observation is trying to be carried over into the area of phylogeny (hypothetical family trees) and whether or not it would be computationally plausible to predict whether or not genes share this kind of relationship. 

 

If genes mutate independent of each other then when given a tree topology, hypothetical ancestors can be predicted in polynomial time.  Without the assistance of the topology (called the large parsimony problem), finding an evolutionary tree in NP-complete. My research has been in regards to a simpler version of the problem: when the tree topology is given, with an added constraint of pairing, does there still exist a polynomial-time algorithm for predicting ancestors?



Short Bio: