Structural variant detection was performed using ABySS (v1.3.2) and trans-ABySS (v1.4.6). For RNA-seq assembly alternate k-mers from k50-k96 were performed using positive strand and ambiguous stand reads as well as negative strand and ambiguous strand reads. The positive and negative strand assemblies were extended where possible, merged and then concatenated together to produce a meta-assembly contig dataset. The genome (WGS) libraries were assembled in single end mode using k-mer values of k24, and k44. The contigs and reads were then reassembled at k64 in single end mode and then finally at k64 in paired end mode. The meta-assemblies were then used as input to the trans-ABySS analysis pipeline (Robertson et al., 2010).
Large scale rearrangements and gene fusions from RNA-seq libraries were identified from contigs that had high confidence GMAP [ref 1] alignments to two distinct genomic regions.  Evidence for the alignments were provided from aligning reads back to the contigs and from aligning reads to genomic coordinates. Events were then filtered on read thresholds. Large scale rearrangements and gene fusions from WGS libraries were identified in a similar way, but using BWA (v0.6.2-r126) alignments.
Insertions and deletions were identified by gapped alignment of contigs to the human reference using GMAP for RNA-seq and BWA for WGS. Confidence in the event was calculated from the alignment of reads back to the event breakpoint in the contigs.  The events were then screened against dbSNP and other variation databases to identify putative novel events.
To determine compartment specific events the structural variant calls for each patient from all matched genome and RNA-seq samples were concatenated together and screened against matching genome tumour, and where available germline bam files. This resulted in compartment specific structural variant events and where germline was available putative somatic and germline events. The events were further filtered against a compendium of germline structural variants to remove recurrent false positives.

[ref 1] Robertson, G., Schein, J., Chiu, R., Corbett, R., Field, M., Jackman, S. D., Mungall, K., Lee, S., Okada, H. M., Qian, J. Q., et al. (2010). De novo assembly and analysis of RNA-seq data. Nature methods 7, 909-912.