Fig. 7
S is a sequenced strain, which contains one transferred segment T from reference sequence R2. B is a HGT breakpoint, which is supported by one paired-end read and two split reads a1 and b2. Through LEMON pipeline, we detect two HGT insertion sites B1 and B2 on references R1 and R2 respectively. For each split read, one portion of it is aligned to the left side of B1 and the other portion is aligned to the right side of B2. We then extract two 100 bp sequences s1 and s2 as input data of DeepHGT. B1 and B2 are in the middle of s1 and s2 respectively