Fig. 1

Classification of reads to find and annotate MGEs. a When reads are aligned with local alignment, there exist three types of reads: reads are mapped (M, Reads 1, 2, 6, and 7); reads spanning MGE insertion are soft-clipped (S, Reads 1, 3, 5, and 7); reads from MGE insertions are unmapped (U, Reads 2, 3, 4, 5, and 6). b All reads are grouped into six classes: class 1 (M-U or U-M), class 2 (S-U or U-S), class 3 (M-S or S-M), class 4 (S-S), class 5 (M-M), and class 6 (U-U). Soft-clipped reads in classes 2, 3 and 5 are used to find breakpoints. The read pairs in classes 1 and 2 are considered as one-end unmapped reads, which are anchored at the upstream or downstream of the breakpoints. These one-end unmapped reads are used to annotate MGEs inserted. c Clipped subsequences (in yellow) at the breakpoints are used to check the integrity of the reads aligned. Breakpoints that are poorly aligned with clipped subsequences are discarded. d The vicinity of the breakpoint is estimated if a few nucleotides are the same. Such situation could occur when the inserted fragment has target site duplication. Read 1 at the beginning of MGEs might be the same as those in the downstream of the breakpoints. Alternatively, read 2 at the end of MGEs might be the same as those in upstream