Logo du LIFL

Depuis le 1er janvier 2015 le LIFL et le LAGIS forment le laboratoire CRIStAL

  1. Actualités

Séminaire de

MICHAL Ziv-Ukelson & Michael Brudno

10 décembre 2010
Salle du conseil de l'INRIA

Seminaire de MICHAL Ziv-Ukelson & Michael Brudno

Michal Ziv-Ukelson (Ben Gurion University) “Computational Aspects of RNA Structural analysis

Though not as abundant in known biological processes as proteins, RNA molecules serve as more than mere intermediaries between DNA and proteins, e.g. as catalytic molecules and gene expression regulators. In the analysis of RNA molecules, structural stability and structural conservation are considered to be key indicators of functionality. This motivated a family of dynamic programming algorithms for RNA secondary structure prediction [Nussinov-1980, Zuker-1981] and for RNA simultaneous alignment and folding [Sankoff-85], which are considered among the few major breakthroughs in the field of bio-molecular structure prediction. The time and space complexities of these algorithms are classically O(n^{3k}) and O(n^{2k}), respectively, where n denotes the maximum length of an input string and k denotes the number of strings to be considered. The "RNA revolution", inspired by recent discoveries of non-coding RNAs, motivates new attempts to improve these algorithms. In my talk I will tell you about some work done in my group to speed up these algorithms, to reduce their space requirements and to increase their accuracy. If time permits, I will also briefly describe some other new bioinformatic tools recently developed in my group.

Michael Brudno (University of Toronto) “Discovering INDEL and Copy Number Genomic Variation from High-Throughput Sequencing

High throughput sequencing (HTS) technologies have enabled the inexpensive sequencing of human genomes, and the discovery of some genomic variants from the resulting short read datasets is well underway. In this talk I will present algorithms for discovery of two types of variants from HTS data: smaller indels (<50bp) and copy number variants (CNVs). First, I will describe MoDIL: Mixture of Distributions Indel Locator, a novel method for finding insertion/deletion polymorphisms from paired short reads. We explicitly model each genomic locus as a mixture of two haplotypes, and our method takes advantage of the high clone coverage to identify both homozygous and heterozygous variation, even if the individual clone sizes are unreliable. Analysis of a recently sequenced genome demonstrates that MoDIL accurately identifies indels >= 20 nucleotides. I will then describe a method to predict CNVs from paired short reads. Our method combines information from paired short reads to identify variable regions and depth-of-coverage to predict the true copy count in the donor genome. Together, the two datasets help overcome both sequencing biases of HTS platforms and spurious read mappings. Our method allows for the detection of CNVs within segmental duplications. We use our method to detect CNVs within the same dataset, and make a total of ~5000 calls that show high concordance with previously known CNVs in this individual.


UMR 8022 - Laboratoire d'Informatique Fondamentale de Lille - Copyright © 2012 Sophie TISON - Crédits & Mentions légales

Page respectant XHTML et CSS.

Pour tout commentaire / Comments and remarks : webmaster