Study Design

• Tissues
• Technology
  • Microarray Production
  • Analysis Techniques
• Network Motifs
  • Network Analysis

• Raw Data
• Literature Evidence of Autoregulation
• Binding Table of Analyzed Data
• Binding Site Presence in Autoregulatory Promoters
• Supporting Material (pdf)
• ChIP Protocol (pdf)

The following describes the design of the 10-slide promoter arrays which contain approximately 400,000 features used in this study. Arrays were produced by Agilent Technologies (www.agilent.com).

To select well-characterized transcription start sites, we first collected the coordinates of all transcription start sites described in five different databases: RefSeq, Ensembl, MGC, VEGA (www.vega.sanger.ac.uk) and Broad (www.broad.mit.edu). The first three are commonly used databases for gene annotation, the last two are manually annotated databases covering subsets of the human genome from the Sanger Institute and Broad Institute, respectively. We then filtered for all transcription start sites that appeared in any two of these five databases (start sites separated by less than 500 bp in any of the databases were considered identical for this step). In cases where there were multiple start sites with different genomic coordinates, we selected the start site that would result in the longest transcript. A total of 18,432 start sites were selected. All sequences and coordinates are from the May 2004 build of the human genome (NCBI build 35), using the repeatmasked (-s) option which separates the genome into masked and unmasked subsequences.

We used the program ArrayOligoSelector (AOS, Bozdech et al., 2004) to score 60-mers for every unmasked subsequence greater than 62 bp across all promoter regions. The scores for each oligo were retained but not put through the built-in AOS selection process.

The collection of scored 60-mers was divided by promoter and sorted by genomic position. Each set of 60-mers was then filtered based on the oligo scoring criteria. AOS uses a scoring system for four criteria: GC content, self-binding, complexity and uniqueness. For our most stringent filter, we selected the following ranges for each parameter: GC content between 30 percent and 100 percent, self-binding score less than 100, complexity score less than or equal to 24, uniqueness greater than or equal to -40.

From this subset of 60-mers, we selected oligos designed to cover the promoter region with an estimated density of one probe every 280 basepairs. To achieve more uniform tiling, we instituted a simple method to find probes within a particular distance from each other. Starting at the upstream end of the region, we selected the first qualified probe, then selected the next qualified probe that was between 150 bp and 280 bp away. If there were multiple, eligible probes, we chose the most distal probe within the 280 bp limit. If there were no probes within this limit, we continued scanning until we found the next acceptable probe. The process was then repeated with the most recently selected probe until we reached the end of the promoter region.

For regions that were not covered by high quality probes, we returned to the full set of scored 60-mers and filtered using a relaxed criteria. This gave us an additional set of 60-mers that we then used to fill gaps in our coverage. After this second pass, we identified gaps in our coverage and added oligos that were properly spaced and best fit our criteria regardless of whether they passed the filter cutoffs. This iterative process gave us a compromise between optimal probe quality and optimal probe spacing. For each start site, we selected the region 8 kb upstream and 2 kb downstream of the site for tiling.

We added several sets of control probes (2,043 total) to the array designs. On each array, there are 40 oligos designed against six Arabidopsis thaliana genes and printed in triplicate. These Arabidopsis oligos have been carried over from previous array designs and were intended for eventual use with spike-in controls. These oligos were BLASTed against the human genome and do not register any significant hits. An additional 543 Arabidopsis oligos were selected as negative controls based on their failure to show any significant BLAST hits against the human genome. Since E2F4 chromatin immunoprecipitations can be accomplished with a wide range of cell types and have provided a convenient positive control for ChIP-Chip experiments (for putative regulators where no prior knowledge of targets exist, for example), we added a total of 80 oligos representing four proximal promoter regions of genes that are known targets of the transcriptional regulator E2F4 (NM_001211, NM_002907, NM_031423, NM_001237). Each of the four promoters is represented by 20 different oligos that are evenly positioned across the region from 3 kb upstream to 2 kb downstream of the transcription start site. We also included a control probe set that provides a means to normalize intensities across multiple slides throughout the entire signal range. There are 384 oligos printed as intensity controls; based on test hybridizations, this set of oligos gives signal intensities that cover the entire dynamic range of the array. Twenty additional intensity controls, representing the entire range of intensities, were selected and printed fifteen times each for an additional 300 control features. We also incorporated 616 "gene desert" controls. To design these probes, we identified intergenic regions of 1 Mb or greater and designed probes in the middle of these regions. These are intended to identify genomic regions that are least likely to be bound by promoter-binding transcriptional regulators (by virtue of their extreme distance from any known gene). We have used these as normalization controls in situations where a factor binds to a large number of promoter regions. In addition to these 2,043 controls, there are 2,256 controls added by Agilent (standard) and a variable number of blank spots bringing the total number of features on each slide to 44,290.

	Start		End
Slide	Chr	Pos	Chr	Pos	Probes
1	1	5575	1	224646230	39961
2	1	224694779	3	108726269	39909
3	3	109290599	5	147564193	39937
4	5	147665548	7	106280884	39935
5	7	106395416	10	15044190	39925
6	10	15119596	11	129697251	39905
7	11	129802259	14	94119500	39930
8	14	94140702	17	41335175	39938
9	17	41603407	20	30042900	39940
10	20	30054185	Y	57685547	39930

Array Designs