When is it advantageous to use a non-splice aware aligner like BowTie2 over a splice aware one like Hisat2?

Question

[pubsurfted]

From what I have read, it is suitable to use non-splice aware aligners for short RNA-seq reads. What I'm having trouble understanding is that: a splice aware aligner can also align those short RNA-seq reads, so why use a non-splice aware aligner? Are there any scenarios in which a non-splice aware aligner will perform better than a splice aware one?

Answer 1

[GenoMax]

aligners for short RNA-seq reads

If you are referring to microRNA (or any other cases where you don't expect the entity to be spliced and the size is going to be small) then you may actually want/need to use an aligner that does not introduce any gaps in alignments.

[colindaven]

Unspliced aligners are also relevant for unspliced mRNA, eg for bacterial RNA-seq.

Answer 2

[i.sudbery]

You would use a non-splice aware aligner, such as Bowtie2 when you are ailgning data that should not have any splicing.

This would be true when aligning reads from technolologies that sequence DNA: WES, WGS, ChIP-seq, ATAC-seq etc.

In these cases you would use a splice unaware aligner like Bowtie2 because they are faster, and have lower memory footprints (STAR has a noteriously high memory footprint). Moreoever, any spliced alignment is likely to be wrong.

You would generally not use a splice unaware tool for mapping RNAseq.

THere is one improtant exception to that: If you are aligning a transcriptome sequence rather than a genome sequence, for further downstream processing by a tool like RSEM.

Here RSEM wants to know that a read came from a particular transcript. If the read aligns with a splice, then it could not have come from that transcript because transcripts in the reference sequence are recorded already spliced.