Figure 4. Detection of fusion transcripts from MAS-ISO-seq of 9 cancer cell lines.

(a) Counts of fusion predictions according to cell line and prediction method, requiring a minimum of 3 long reads as supporting evidence. Line drawn indicates the number of fusions agreed upon by at least two methods. (b) Numbers of MAS-ISO-seq reads identified as evidence for COSMIC fusions according to method. (c) Fusion transcript detection accuracy according to minimum long reads supporting evidence based on the proxy truth set. (d) Comparison of long (MAS-ISO-seq) vs. short read (TruSeq Illumina) support for fusion isoforms detected by each according to CTAT-LR-fusion and FusionInspector, respectively. Read support is normalized for sequencing depth as FFPM. (e, f) Five fusion isoforms observed for the fusion gene CYTH1::EIF3H of cell line SKBR3 are (e) observed with highly correlated expression measurements as estimated from long and short RNA-seq reads and (f) shown according to fusion transcript breakpoints.