search

lh3 / minimap2

A versatile pairwise aligner for genomic and spliced nucleotide sequences
https://lh3.github.io/minimap2
Other
1.8k stars 412 forks source link

MiniMap2 and splice site detection #108

Closed DepledgeLab closed 6 years ago

DepledgeLab commented 6 years ago

Hi Heng,

I'm not clear how minimap2 is functioning in terms of identifying splice sites. Given that the sequence reads may contain base miscall and/or indel type errors, it seems likely that splice acceptor and donor recognition may be compromised in some positions?

I then started wondering whether minimap2 looks at the reference genome/transcriptome or the sequence read itself to determine whether a splice donor/acceptor is utilised in a given situation (and indeed whether it favours a canonical or non-canonical donor/acceptor sequence)?

lh3 commented 6 years ago

If there are no errors or ambiguity, naive alignment directly gives you splice junctions. In more complex cases, minimap2 as well as most other RNA-seq mappers prefers canonical signals. See the minimap2 preprint for details.