annuncio seminario

[Alessandra Guglielmi <alessan@mi.imati.cnr.it>]

Il giorno 19 novembre 2004, alle h 10.00, si terra' presso il 
CNR-IMATI, sezione di Milano, Via Bassini, 15, Milano, nella 
sala riunioni (secondo piano), il seminario

        MODEL-BASED ESTIMATION OF TRANSCRIPT CONCENTRATIONS 
                   FROM SPOTTED MICROARRAY DATA

                         ARNOLDO FRIGESSI

                     Department of Statistics 
	        Institute of Basic Medical Sciences 
	               University of Oslo
                               and 
		   Norwegian Computing Centre 
		              Norway
		    

Abstract.

After an introduction to the challenges posed to statistics by microarray
data and the related scientific questions, we present a new method which
provides estimates of absolute transcript levels from spotted microarray data
based on a hierarchical Bayesian model for the entire microarray experiment.
Transcript levels are universal and can be included in further analysis with
similar estimates from disparate experiments in other laboratories
(meta-analysis). Our method is flexible with respect to input data; it does
not require a common reference or any special experimental design, but some
genes spotted in replicates are needed. The hierarchical model utilizes
information on all steps in the experiment, included as covariates in a model
for the probability for an mRNA molecule in the target solution to 'survive'
the journey until it is hybridized and contributes to the measured intensity
in a certain pixel. The image analysis part of the experiment is integrated
in the hierarchical model, and based on pixel-wise intensity measures; the
Bayesian regime via MCMC allows backward calculation of the underlying
unknown transcript levels. The Bayesian model provides the joint distribution
of all transcripts, which suggests several new alternatives to gene
selection. We validate the estimation of transcripts levels in an experiment
with known concentrations (spikes), and we explore our new statistical tools
for data analysis on a small dataset with samples from a human cervical
tumor.

This is joint work with: Mark A. van de Wiel, Department of Mathematics and
Computer Science, Technische Universiteit Eindhoven, The Netherlands; 
M. Holden, Norwegian Computing Center, Oslo, Norway; Ingrid K. Glad, 
Department of Mathematics, University of Oslo, Norway; Heidi Lyng, 
Department of Biophysics, The Norwegian Radium Hospital, Oslo, Norway.



Tutti gli interessati sono cordialmente invitati a partecipare.
-- 
Alessandra Guglielmi             e-mail: alessan@mi.imati.cnr.it
CNR-IMATI                        tel.  : ++39.02.23699529 
via Bassini, 15                  fax   : ++39.02.23699538 
20133 Milano (ITALIA)

=============================================================================
NOTA:
Le norme per utilizzare il forum SIS e le istruzioni per iscrizione
e cancellazione sono disponibili all'indirizzo
.
.   http://w3.uniroma1.it/sis/forum.asp
.
L'archivio di tutti i messaggi (aggiornato al giorno precedente)
e' disponibile all'indirizzo
.
.   http://www.stat.unipg.it/cgi-bin/wilma/sis/
.
=============================================================================