Fig. 4

Identification of gene fusion events in cancer cell lines by both RNA-seq and WGS. a Out of 5277 candidate fusion events identified by RNA-seq analysis (using STAR-fusion algorithm) involving 327 cell lines with WGS data [10], numbers of events with support from somatic SV analyses. As indicated, for 2307 candidate fusion events, SV breakpoints were found within one or both genes, with and without a high expression association. A high expression association is defined here as one of the following: (1) for fusion events occurring in one or two tumors, whether for each tumor the expression of either gene was > 0.4SD from the median; or (2) whether a significant association between SV breakpoints and increased expression (p < 0.01, linear model incorporating tumor type and CNA) was found for either gene, either by distance metric method or by genomic region window within the gene body. b Of the 853 gene body SV breakpoint events associated with overexpressed genes (over-expression defined as > 0.4SD from the sample median, events involving the set of 192 genes from Fig. 2a with FDR < 10%, correcting for tumor type and CNA), the fractions of events associated with either gene fusion by combined RNA-seq and SV analysis or high-level gene amplification are indicated. c Gene fusions with both RNA-seq and SV support (i.e., breakpoints detected for at least one of the two genes) with high expression association (part a) and involving more than two cell lines are represented. Cancer type is indicated along the top and in the coloring of the fusion event