Novel piRNA processing pattern involving short by-products of piRNA targeting. (A) Sketch illustrating how the 5' distance of sequences from opposite strands was measured. (B) Frequencies with which various distances between the 5' ends of sequences derived from opposite chromosome strands occur in the mouse testes lysate data. On the x-axis, the 5' offset of sequences deriving from opposite strands is shown, and on the y-axis, the number of detected 'pairs'. The analysis was carried out for several subsets of sequences defined by the length of reads taken into account: blue line - all sequences of length 15-35 nt, red line - only sequences in the range of prototypical piRNAs (23-32 nt), green line - pairs were only counted if they involved on one strand a sequence in the range of piRNAs (23-32 nt) and on the opposite strand a sequence below that range (15-22 nt), black - as for green except that the short sequence had to be precisely 19 nt long. The peak at 9 (P9) is generated by sequences in the piRNA range, while the peak at 28 (P28) comes from pairs in which one sequence is long and the other is short, predominantly 19 nt. (C) Duplexes corresponding to the P9 (top) and P28 (bottom) patterns. The centers of the duplexes, which are used to analyze the co-occurrence of the two patterns are colored in blue and green respectively. (D) Histogram of the length distributions for the long (23-32 nt) and short (15-22 nt) sequences that contribute to the P28 pattern.