Fig. 3

FuSpot Fusion Alignment Algorithm Visualized. A conceptual representation of the FuSpot alignment algorithm. Each colored line represents an overhead view of a 2D Smith Waterman score matrix representative of a given read and one of eight reference sequences (4 on the 5′ end of the breakpoint and 4 on the 3′ end.) The central black circle marks the fusion breakpoint. Since fusion analysis is typically conducted with RNA data, the ability to align against many references simultaneously is crucial. Here the chosen references exemplify how FuSpot can be used to align a set of reads, each of which may be fusion or non-fusion as well as genomic or exonic. During FuSpot realignment, should backtracking commence from a matrix on the 3′ side of the breakpoint, tracing could follow through to the breakpoint. In such a case, FuSpot’s realignment algorithm would search the rightmost column of each of the four matrices on the 5′ end of the breakpoint for the appropriate next step to trace. Once determined, it will follow the dotted line to that matrix, then trace through it for the remainder of the backtrack. Subsequently, the traced 5′ and 3′ matrices and their associated references will be assigned a color during FuSpot’s visualization step and the aligned read will take on the appropriate color for the reference to which it aligned on either side of the breakpoint