Fig. 1

Affinity-seq determines genome-wide PRDM9 binding sites. a Shape of a representative Affinity-seq peak. The size and position of the inferred PRDM9 binding site are shown on top; the size and position of the corresponding H3K4me3 hotspot are shown on the bottom. b Comparison between the composite signals (obtained by aggregating across all hotspots in the genome) at H3K4me3 and Affinity-seq peaks centered on the inferred PRDM9 binding sites. c Distribution of nucleotide frequencies along PRDM9 binding sites including the flanking nucleotides. Distinct preferences can be seen over the entire binding sites; the strongest signals are detected at positions covered by the PRDM9^Dom2 motif indicated on top (nucleotides 7–19). The sequence of the binding motif identified in the nucleosome-depleted regions of H3K4me3 peaks and the structure of the zinc finger domain of PRDM9^Dom2 are shown above. Note that, except for the terminal fingers, nucleotide frequencies are distinctly different from genome average for all positions, including those outside the region of the computationally derived motif; that identical fingers (QDK, QVK and especially AVQ) vary in nucleotide frequencies depending on their location in the array, and that nucleotide frequencies at the same amino acid position (notably Q in fingers 3 and 4 v. 5 and 6) are strongly influenced by the identity of adjoining amino acids