TopHat2, Accurate Alignment of Transcriptomes in the Presence of Insertions, Deletions and Gene Fusions

Link from Genome Biology h/t: @genetics_blog

The problem:

More importantly for the alignment problem, ~20% of junction-spanning reads extend 10 bp or less into one of the exons they span. These small anchors make it extremely difficult for alignment software to map reads accurately, particularly if the algorithm relies (as most do) on an initial mapping of fixed-length k-mers to the genome. This initial mapping, using exact matches of k-mers, is critical for narrowing down the search space into small local regions where a read is likely to align. If a read only extends a few bases into one of two adjacent exons, then it often happens that the read will align equally well, but incorrectly, with the sequence of the intervening intron. For example, as illustrated in Figure 1, suppose that read r spans exons e1 and e2, extending only 4 bases into e2. Suppose also that that e2 begins with GTXX, and the intervening intron also begins with GTXX, where X stands for any of A, G, C, and T. Then r might align perfectly to e1 and the first 4 bases of the intron, and the alignment algorithm will fail to find the spliced alignment of r.

Solution:

In order to handle this problem, TopHat2 uses a two-step procedure. First, similar to TopHat1 [4], it detects potential splice sites for introns (detailed further in Materials and methods). It uses these candidate splice sites to align multi-exon spanning reads properly in a subsequent step. Some RNA-seq aligners, including GSNAP [5], RUM [6], and STAR [7], map reads independently of the alignments of other reads, which may explain their lower sensitivity for these spliced reads (see Results and discussion). MapSplice [8] uses a two-step approach similar to TopHat2.

When you have good annotation, such as in human genome, TopHat2 can also use the annotations “during its initial mapping phase, which produces significant gains in sensitivity and accuracy”.

Finally, color space is included !

The paper has many good comparisons with other alignment programs. For example,

GSNAP, RUM, and STAR have particular difficulty aligning short-anchored reads, only aligning 26%, 8.6%, and 3.5%, respectively. MapSplice does considerably better, aligning 75.6% of these reads. By contrast, TopHat2 aligns 93.7% of the short-anchored reads using Bowtie1 as its main aligner (Table 1).

Overall TopHat2 is a big improvement over a program that was excellent itself.