Embryos were collected aged appropriately, then dechorionated and formaldehyde fixed. Fixed embryos were then homogenized, nuclei were pelleted, and chromatin was sheared by sonication.

Chromatin was immunoprecipitated using the appropriate antibody and the recovered material was amplified by PCR.

The recovered material was characterized by Affymetrix whole-genome tiling array.

Data analysis was as described elsewhere (FBrf0205197). ChIP-chip array data were processed using TiMAT. Only data for oligos mapping uniquely to release 4 of the D. melanogaster genome were considered. The mean hybridization intensity at each probe was divided by the mean probe intensity in the input DNA samples (three technical replicates). The logarithms of the ratios (IP/input) were averaged in 675-bp windows advanced one oligo at a time (lowest and highest values were dropped to produce a trimmed mean). Bound regions were identified by comparing the distribution of observed window scores to the computed symmetric null distribution: see Figure 1F in Li et al. (FBrf0205197). This estimated null distribution was used to assign p-values to each window, and a score threshold was chosen at 1%FDR. Bound regions ("intervals") were calculated from contiguous stretches of windows having scores above the given FDR threshold, requiring these contiguous stretches to contain at least ten windows with a maximum allowable gap of 200bp between any two adjacent windows.