Hi, I am involved in the identification of new breast tumor associated antigens (TAA) suitable for molecular imaging and vaccination. The main issue, for the kind of TAAs, is their relative low expression in normal tissue and high expression in tumor specimens. Although I have made some in house microarray experiments on tumor specimens I would like to take advantage of breast cancer studies and mammary gland normal tissue, both available in GEO. The question I would like to address is the identification of genes that behave differently between normal and tumor specimens (low expression in normal samples and high in tumor specimens) but the main point is that the two experiments were performed in different labs, different samples, different preparation procedures, etc. I read the paper from Gentleman et al. On the Synthesis of Microarray Experiments <http: www.bepress.com="" bioconductor="" paper8=""> (September 30, 2005) which suggests the use of Random Error Model (REM) to integrate different studies. However, my situation is different from the conventional data integration where two different studies, e.g. comprising different tumor histotypes, need to be integrated. In my case the comparison is made between normal and tumor samples coming from different studies. Therefore, it is not clear to me if modeling normal and tumor samples using the REM approach proposed in Gentleman will be helpful also in the identification of genes that behave differently in normal and tumor specimens. I will be very pleased to receive any comment and suggestion. Raffaele ---------------------------------------- Prof. Raffaele A. Calogero Bioinformatics and Genomics Unit Dipartimento di Scienze Cliniche e Biologiche c/o Az. Ospedaliera S. Luigi Regione Gonzole 10, Orbassano 10043 Torino tel. ++39 0116705420 Lab. ++39 0116705408 Fax ++39 0119038639 Mobile ++39 3333827080 email: raffaele.calogero at unito.it www: www.bioinformatica.unito.it