TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions

Table 1 Performance of TopHat2 and other spliced aligners on a set of 20 million 100-bp, single-end reads, simulated based on transcripts from the entire human genome.

Program	No. of mapped reads	Correctly mapped reads, %	Incorrectly mapped reads, %	Unmapped reads, %	Correct junction reads, %^a	Correct short-anchored reads, %^b
TopHat2 + Bowtie1	19,826,638	98.31	0.82	0.87	95.28	93.69
TopHat2 + Bowtie2	19,826,673	98.03	1.10	0.87	94.28	89.67
TopHat1.14	19,616,874	94.64	3.45	1.91	84.44	44.08
GSNAP	19,997,255	94.21	5.77	0.02	83.15	26.01
RUM	19,555,823	88.11	9.67	2.22	65.35	8.59
MapSplice	19,872,372	97.28	2.08	0.64	92.09	75.57
STAR	19,087,508	92.14	3.30	4.56	77.17	3.54

^aThere were 6,862,278 reads spanning one or more splice junctions; the alignment accuracy of junction reads refers to this set.
^bThere were 1,448,022 reads extending 10 bp or less into one exon; the alignment accuracy of the short-anchored reads is based on these alignments.

ISSN: 1474-760X