Universität Bielefeld - Technische Fakultät - AG Genominformatik

Genominformatik

AG-Seminar im Sommersemester 2003

Donnerstag, 16-18 Uhr, V4-106

Jens Stoye, Jomuna V. Choudhuri

---

Kurzbeschreibung

In dieser Veranstaltung wird in Vorträgen über aktuelle Themen aus der Forschung der Arbeitsgruppe Genominformatik berichtet.

Zeitplan

Datum	Referent	Titel
23.10.2003	Jomuna V. Choudhuri	GenAlyzer - Overview and Biological Applications
30.10.2003	diverse	RNA WORKSHOP
6.11.2003	Jens Stoye	On Genomic Distances
13.11.2003	Claudia Fried	Footprinting: trailing conserved noncoding sequences in the genome
20.11.2003	Ferdinando Cicalese	Threshold Group Testing
27.11.2003	Thomas Schmidt	Finding all gene clusters in 2 sequences
4.12.2003	-	-
11.12.2003	Klaus Schürmann	A fast construction algorithm for suffix arrays
18.12.2003	Manuel Scholz	Challenges in Mobile Ad Hoc Networks
25.12.2003	X-MAS	-
1.1.2004	NEW YEAR	-
8.1.2004	Zsuzsanna Liptak	Finding submasses in weighted sequences with FFT
15.1.2004	Constantin Bannert	Introduction to PASSTA
22.1.2004	Patrick May	PTGL - Protein Topology Graph Library
29.1.2004	Sebastian Böcker	Weighted Sequencing from Compomers and beyond...
5.2.2004	Michael Spitzer	RiPE: Sampling low-conserved regions for homology-search based tree reconstruction
12.2.2004	Michael Kaltenbach	?
19.2.2004	Rileen Sinha	?
	Kim Rasmussen	?
26.2.2004	Gregor Obernosterer	Identifying local stable sRNA structures for genome wide surveys
	Thomas Schmidt	?

Title: "Threshold Group Testing"
Speaker: Ferdinando Cicalese


Dipartimento di Informatica ed Applicazioni "R.M. Capocelli"
Universita' egli Studi di Salerno, ITALY
20.11.2003, 16:00 c.t., V4-106

Abstract
--------

The classical group testing problem is described as follows. In a set U of n elements, p elements are "positive" and the other n-p are "negative". A group test consists in asking whether a specified subset S of U contains at least one of the positive elements. The goal is to identify the set P of all positive elements by using the least possible number of group tests. Group testing is of interest in chemical and biological testing, DNA mapping, and also in several computer science applications. We introduce a natural generalization of the group testing problem: A test gives a positive (negative) answer if the pool contains at least $u$ (at most $l$) positive elements, and an arbitrary answer if the number of positive elements is between these fixed thresholds $l$ and $u$. We show that the positive elements can be determined up to a constant number of miscalssifications, bounded by the gap between the threeesholds. This is in a sense the best possible outcome. We also study the number of tests needed to achieve this goal. We show that in special cases Threshold Group Testing can be as efficient as Classical Group Testing .
--------

Title: "Finding all gene clusters in 2 sequences"
Speaker: Thomas Schmidt


27.11.2003, 16:00 c.t., V4-106

Abstract
--------

A popular approach in comparative genomics is to locate groups or clusters of orthologous genes in multiple genomes and to postulate functional association between the genes contained in such clusters. To this end, genomes are often represented as permutations of their genes and common intervals, i.e., intervals containing the same set of genes, are interpreted as gene clusters. A disadvantage of modelling genomes as permutations is that paralogous copies of the same gene inside one genome can not be modelled. In the talk I consider a slightly modified model that allows paralogs, simply by representing genomes as sequences rather than permutations of genes. I will sketch a simple algorithm that finds all common intervals of two genomes in O(n²) time using O(n²) space and afterwards I will show, more complicated algorithm runs in O(n²) time and uses only linear space.
--------

Title: "Challenges in Mobile Ad Hoc Networks"
Speaker: Manuel Scholz


17.12.2003, 16:00 c.t., V4-106

Abstract
--------

The aspect of mobility has gained much attention in today's research in computer science. Especially mobile ad hoc networks are the focus of many research groups. This talk will give a short introduction and survey of the possibilities and the challenges of mobile ad hoc networks. Furthermore it will introduce patterns of cooperation among the network clients and will describe the problem of data dissemination inside these networks in detail.
--------

Title: "PTGL - Protein Topology Graph Library"
Speaker: Patrick May


22.01.2004, 16:00 c.t., V4-106

Abstract
--------

The experimental exploration of protein structures is a basic topic of post-genomic research. Therefore, the theoretical analysis and description of protein structures to search for similar protein structures became more and more important. PTGL is a database for protein topologies. The arrangements of secondary structure elements (SSEs) to form motifs and domains in protein structures opens the possibility of their topological description. Topological representations of protein structures are based on sequences of its SSEs, i.e., helices and strands. At this description level protein topologies are defined as undirected labelled graphs. The simplest representation of protein topology are schematic diagrams of protein folds illustrating the SSEs and their spatial neighbourhood. We present a mathematically unique linear notation with a new type of schematic representation, which we have implemented in PTGL combined with an online search tool for data interrogation by sequence similarity and keyword queries.
--------

Title: "RiPE: Sampling low-conserved regions for homology-search based tree reconstruction"
Speaker: Michael Spitzer


05.02.2004, 16:00 c.t., V4-106

Abstract
--------

The homology search stage of RiPE (Retrieval induced Phylogeny Environment) has been significantly improved by enhanced sampling of low-conserved regions. This is done by introducing an additional iteration searching for low- conserved regions only and incorporating the results into the search profile. The resulting homologous sequence fragments are larger, resulting in a better supported and more complete tree. Secondly, the isoform recognition is improved by using a trained support vector machine instead of fixed thresholds.
--------