Table 4.

Comparisons of results from using single-end and paired-end reads for finding fusions

Read type	Fusion genes (left-right)	Chromosomes (left-right)	5' position	3' position	Spanning reads (RPM)	Spanning pairs
Single	GAS6-RASA3	13-13	114529968	114751268	15 (0.267)
Paired	GAS6-RASA3	13-13	114529968	114751268	10 (0.198)	43
Single	BCR-ABL1	22-9	23632599	133655755	6 (0.107)
Single	BCR-ABL1	22-9	23632599	133729450	3 (0.053)
Paired	BCR-ABL1	22-9	23632599	133655755	2 (0.040)	7
Paired	BCR-ABL1	22-9	23632599	133729450	3 (0.059)	10
Single	ARFGEF2-SULF2	20-20	47538548	46365683	17 (0.302)
Paired	ARFGEF2-SULF2	20-20	47538545	46365686	10 (0.198)	30
Single	BCAS4-BCAS3	20-17	49411707	59445685	25 (0.445)
Paired	BCAS4-BCAS3	20-17	49411707	59445685	13 (0.257)	145

Comparisons of single-end and paired-end reads as evidence for gene fusions in the Universal Human Reference (UHR) cell line (a mixture of multiple cancer cell lines), using the known fusions GAS6-RASA3, BCR-ABL1, ARFGEF2-SULF2, and BCAS4-BCAS3. With TopHat-Fusion's ability to align a read across a fusion, the single-end approach is competitive with the paired-end-based approach. RPM is the number of reads that span a fusion per millon reads sequenced. For instance, the RPM of single-end reads in GAS6-RASA3 is 0.267, which is slightly better than the RPM for paired-end reads. Single-end reads may show higher RPM values than paired-ends in part because single-end reads are longer (100 bp) than paired-end reads (50 bp) in these data, and therefore they are more likely to span fusions.