The dorsal patterning in Drosophila is controlled by an extracellular gradient of the morphogens Decapentlaplegic/Screw (Dpp/Scrw), which are members of BMP/TGF-β family. Dpp/Scrw signal is transduced to the nucleus by the transcription factors, Mad/Medea. The transcriptional effectors exert their regulation in a graded-manner eliciting at least three threshold responses: high, intermediate and low. However, the mechanism underlying differential response to Dpp is poorly understood, due in part to the insufficient number of well-studied target genes. Gene expression changes were analysed in ectopic overexpressing Dpp mutant embryos to identify new target genes of Dpp/Mad pathway. Overall design: Total RNA from nos-Gal4>lacZ and nos-Gal4>Uas-dpp embryos (N=100-200) at stages 5 to 7 of development was extracted as described in Zuñiga et al. (BMC Biology. 7:61, 2009). The fluorophores Cy3 or Cy5 (Amersham) were coupled with aminoallyl-modified nucleotides after amplification of total RNA following manufacturer’s instructions (MessageAmp II aRNA Amplification Kit (Ambion, cat. AM1753). After purification, dye incorporation was calculated. Experimental and control labeled 2.5 μg of aRNA probes were co-hybridized to microarrays in a blocking solution. Glass slides with probe spots distributed in 48 blocks representing the Drosophila transcriptome were purchased from Microarrays Inc (http://www.microarrays.com/). Two independent control (nos-Gal4>lacZ) and experimental (nos-Gal4>UAS-dpp) embryos samples (biological replicates) were hybridized in nine slides to an oligonucleotide microarray (Microarray Inc.). The experimental and control samples were tagged with Cy5 and Cy3, respectively, three hybridizations were dye swapped. Slides were pre-hybridized, hybridized and washed according to recommendations of Microarray Inc. Images were processed using the software ScanArray Express (Perkin Elmer) to aligning both channels at different PTM gain. Image quality was assessed by descriptors included within the R function, q.com. Additionally, control spots (spike controls, empty spots and random oligomers) were analyzed separately to discriminate high quality hybridized slides. The data from selected slides were processed to remove noise with normexp, normalized by block with loess and to identify differentially expressed genes by using the statistical R package LIMMA. LIMMA is able to estimates the coefficients of the contrast matrix between the conditions and fits a linear model for each gene. Estimation of significance is based on a hypothesis test that uses the statistical t-moderate associated to p-value. P values were adjusted by the method of Benjamini and Hochberg to control the rate of false values. Genes with statistically significant changes in expression (adjusted P-value < 0.05) were selected.