Figure 4.

Short matches define the limits of a deletion. (a) Deletion via replication slippage. Illustrated is a process by which a replication fork slips from the sequence and reconnects at a different locus, thereby deleting the short sequence shown. The stability of a slippage event is controlled by the length of matching sequence (red) at the deletion junction (arrow) (b) Similarity between deleted/inserted sequence and its flanking nucleotides. We computed the similarity between deleted (first row) and inserted (second row) nucleotides and the nucleotides flanking them (comparing each position to the position l nucleotides away, where l is the indel length). We used the flanking sequence in the side with better overall percent identity and averaged the statistics over all optimal alignments to control for alignment algorithm artifacts (see Materials and methods). For insertions we observe high similarity that is unaffected by the distance from the junction, while for deletions the similarity is rapidly decreasing as a function of the distance. D. Sec, D. sechelia; D. Sim, D. simulans. (c) Deletion rates. Shown are deletion rates, as a function of the deletion length, for various junction match lengths (denoted s') from 0 (bottom) to 6 (top). The rates are normalized by the background genomic frequency of identical sequences of length s spaced by l bps. The rate increases by a factor of 100-fold with s, but retains the same slope, regardless of the deletion length l. Combined human and chimp data are shown.