Figure 2

TE-Tracker main algorithms. a. Discordant pairs around insertion breakpoint. Sequenced reads around a newly inserted TE-copy (top half) produce discordant read mappings when aligned onto the reference sequence where the newly inserted copy only exists at the locus of origin (bottom half). The thin black line represents the sequenced DNA fragment, the thick black line represents a transposon of interest. Yellow and orange arrows represent the left and right extremities of the insertion breakpoint, linked arrows represent paired-end reads. Grey reads will be normally mapped, while colored reads will be mapped discordantly, the color indicates a type of discordance (left mate on the acceptor and right on the donor and vice-versa). b. Clustering of discordant pairs. Discordant reads of the same type are isolated and sorted (left half). Both ends must be sufficiently close for two read pairs to be clustered together, but sorting of the left end, combined with a random insert size results in different thresholds for clustering both ends. Pairs are clustered according to the Single-Linkage method (see Methods), which represent read pairs as edges on a graph (right half). A point is added to a cluster if its distance to any other point already in the graph meets both thresholds when projected on both axes. c. Cluster merging. Local drops in read coverage break clusters, corrupting insertion signals. A proximity threshold is applied to merge neighboring clusters of the same type and orientation. Local coverage is represented by a grey curve on top of the sequence, while linked colored arrows represent clusters of read pairs. d. Calling. The four types of transposition events detected by TE-Tracker along with their associated cluster signatures, with an emphasis on the overlap condition used to assemble clusters with compatible signatures into bona fide events.