For instance, in recent studies using the modENCODE [ 42] and ENCODE [ 50, 51] datasets, only the top 250 and 500 binding peaks with a length of 100 bp and 200 bp, respectively, in each dataset were used to find motifs.
We started with training the top 2000 ERα binding peaks with high scores (enrichments) by the ChIPMotifs, a total of 7 TFs were identified, including ERE, PAX6, PITX2, RORA, XBP1, PPARG and PPARA.
Sequences from 100 binding peaks with the greatest number of good probes were chosen from our Stat3 and c-Myc dataset and imported into Weeder.
Binding peaks with FDR < 0.01 were defined as the ARF6 binding peak and used in further analyses.
Using FindPeaks Homer software, Pou3f1 binding peaks with fourfold greater normalized tags were identified in ChIP experiments compared with the control (Heinz et al., 2010).
We next overlapped differentially expressed genes with binding peaks with our initial expression scatterplot (Fig 6A and Supplementary Fig S8) to exclude expression changes due to indirect effects.
ChIP-Seq analysis R (CSAR) software was used to identify binding peaks, with parameters (backg = 10, norm = −1, test = 'Ratio', times = 1e6, digits = 2) (Muino et al., 2011 ).
Binding peaks with FDR <0.01 were finally defined as the ARF6 binding peak and genes having at least one ARF6 binding peak within its promoter (−3 kb) or coding region or 1 kb downstream from stop codon were considered direct target genes.
Still, the large majority of TF binding differences are not associated with genetic changes during evolution to the directly bound sequence motifs, and shared TF binding peaks with conserved intensity (discussed below) were more likely to show depletion of nucleotide substitutions and heightened sequence constraint.
We identified a total of 29,221 HOXC9-binding peaks with FDR less than 1%.
