Whole-genome DamID binding profiles of 53 chromatin proteins in Drosophila Kc167 cells were generated and/or analyzed. In the same array platform, ChIP-on-chip profiles of histone H3, H1, H3K9me2, H3K27me3, H3K4me2, and H3K79me3 were obtained. These were correlated with gene expression, which was measured by RNA-tag profiling.

Cells were grown in BPYE medium with 25% bactopeptone, 20% yeast extract and 5% heat inactivated fetal calf serum. Expression plasmids for Dam-fusion and Dam-only proteins were transfected in parallel by electroporation, and cells were isolated 24 hr later. The Dam enzyme deposits stable adenine-methylation footprings at genomic sites bound by the Dam-fusion.

Genomic DNA was isolated and adenine-methylated fragments were amplified by methylation-specific PCR using Cy-dye labeled random nonamers. To correct for nonspecific binding of Dam and local differences in DNA accessibility, methylated fragments of Kc cells transfected with a Dam-only construct were labeled with a different fluorescent dye.

Isolated DNA was hybridized to NimbleGen Drosophila 385k array rel4 300bp genomic tiling array.

Quantitative DamID genome profiles for 53 chromatin proteins were analyzed to identify distinct combinations of proteins that recur in the genome. Principal component analysis was initially performed to reduce the dimensionality of the data. The first three principal components were then used, which together account for 57.7% of total variance. Doing so, five distinct lobes could be observed in the three-dimensional scatter plot. The result was robust to different quantification methods. Having established that classification into five types properly summarizes the data, a five-state hidden Markov model was fitted onto the first three principal components. Thus, every probed sequence in the genome was assigned one of five exclusive chromatin types. Each of the five chromatin signatures was labeled with a color, and further characterzied by histone modification and gene expression patterns.