Representation of Signal Sets for the Analysis of Pathogen Variability

Speaker: Professor Paul Dan Cristea
Biomedical Engineering Center,
University "Politehnica" of BucharestAbstract: It is often desirable to analyze sets of related signals in a way that separates the joint variation of the set, from the individual features of each component signal. The specific case that prompted our research is the genomic signal analysis of taxon variability, but the problem is general and the methods presented in the paper can be used to improve the description of many other types of signal sets. To better express variations of related signals in a given set , where the individual signals are , with xi – the sampling points and yi – the corresponding signal values, it is advantageous to describe the signal set by using two types of components:
(1) a reference sequence that gives the central tendency of the set, i.e., the common variation
of the signals;
(2) the individual offsets of the signals with respect to the chosen reference:
.
The central tendency that is used as reference must extract optimally the common variation of the signals in the considered set, so that the individual offset of each signal expresses only the features specific to that signal, with minimum crosstalk. Various estimates of the central trend have been used as reference: average sequence of the signals (or other linear combination of the signals), median sequence , mode step sequence , and maximum flat sequence .The method has been used to analyze pathogen variability resulting in treatment resistance. Mutations occurring at specific sites along a segment comprising codons 509-595 of the rpoB gene of Mycobacterium tuberculosis (MT), which encodes thesub-unit of RNA polymerase, can be put in direct correspondence to Rifampin (RMP) resistance. Similarly, the variability of polymerase and reverse transcriptase genes of HIV-1, Clade F, has also been monitored this way to detect resistance to the antiretroviral treatment of AIDS.Biography: Professor Paul Cristea graduated from the Faculty of Electronics and Telecommunications (UPB - University "Politehnica" of Bucharest, Romania, 1962), the Faculty of Physics (University of Bucharest, 1969), and has a Ph.D. in Technical Physics (UPB, 1970). Since then his research and teaching activities covered an large area of Electrical Engineering and related domains including topics like Digital Signal and Image Processing, Genomic Signals, Neural and Evolutionary Systems, Computerized Medical Equipment, Evolutionary Intelligent Agents, Intelligent e-Learning Environments. He is the author or co-author of more then 130 published papers, 11 patents and contributed to more then 20 books in these fields. He is currently affiliated with UPB, the Biomedical Engineering Center (general director) and the Vrije Universiteit Brussel, Belgium, the ETRO Department. He is a corresponding member of the Romanian Academy and director of the Romanian Bioinformatics Society.Host: Prof. Sanjit Mitra

Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248

Audiences: Everyone Is Invited

Contact: Talyia Veal