CSC Weekly Seminar
Friday, January 18, 2013
TASC-2, Rm 8500
2:30 pm

Speaker: Antoine Baker
Affiliation: SFU Physics


Title: Linking the DNA strand asymmetry to the spatio-temporal replication program:
from theory to the analysis of genomic and epigenetic data

Abstract: Twokey cellular processes, namely transcription and replication, require theopening of the DNA double helix and act differently on the two DNA strands, generating different mutational patterns (mutational asymmetry) that may result, after long evolutionary time, in different nucleotide compositions on the two DNA strands (compositional asymmetry). Here, we propose to modelthe spatio-temporal program of DNA replication and its impact on the DNA sequence evolution. The mutational and compositional asymmetries observed in the human genome are shown to decompose into transcription- and replication-associated components. The replication-associated asymmetry is related to thereplication fork polarity, which is also shown to be proportional to the derivative of the mean replication timing. The large-scale variation of the replication fork polarity delineate Mbp scale replication domains where the replication timing is shaped as a U. Such replication domains are also observed in the germline, where they are revealed by a N-shaped compositional asymmetry, which indicates the conservation of this replication program over several hundred million years. The replication domains borders are enriched in open chromatin markers, and correspond to regions permissive to transcription and replication initiation. The analysis of chromatin interaction data suggests that these replication domains correspond to self-interacting chromatin structural units, at the heart of a highly parallelized organization of the replication program in the human genome.

CSC Weekly Seminar
Friday, February 1, 2013
TASC-2, RM 8500
2:30 pm

Speaker: David Bryant, Department of Mathematics and Statistics, University of Otago

Title: How to lasso quickly, positively, and correctly.

Abstract:

The Lasso (also known as L1 regularization) is a method which has become especially popular in statistics because of its ability to automatically select subsets of variables. The LARS-Lasso algorithm of Efron et al. (2004) makes the
lasso computational feasible: with little added computational cost one can construct lasso solutions for all ranges of the shrinkage parameter. In many applications, however, model considerations impose a non-negativity constraint on the variable coefficients, giving a demand for a "positive lasso". Efron et al (2004) propose a LARS type
algorithm for the positive lasso, but it is not guaranteed to produce optimalsolutions. Here we describe a modification to the LARS algorithm which doesproduce optimal positive lasso solutions, and illustrate the algorithm using an example from evolutionary biology.

PIMS-CSC Distinguished Speaker Series

IRMACS Theatre & Atrium
3:30 pm

Speaker: Prof. Mark W. Newman, School of Information, University of Michigan

Title: Epidemics, Erdos numbers, and the Internet:The new science of networks


tba

PIMS-CSC Distinguished Speaker Series

IRMACS Theatre & Atrium
3:30 pm

Speaker: Dr. Tao Tang, Chair Professor, Department of Mathematics, Hong Kong Baptist University

Title: High-Order and Adaptive Time Stepping Methods for Energy Gradient Flows