What is Canonical/Non-canonical splicing?

Question

[gprashant17]

I am using STAR aligner to align an RNA-seq data with a human reference genome and I was analysing the log.final.out file. I am not sure what the following results mean:

Number of splices: GT/AG

Number of splices: GC/AG

Number of splices: AT/AC

Number of splices: Non-canonical :

I understand that the number of splices (total) gives the total number of splices in all the reads and number of splices (annotated) gives the those which are present in the annotation file.

What does canonical/non-canonical splicing mean?

Answer 1

[Jean-Karim Heriche]

The canonical splice sites are those originally described and most commonly found (like in ~99% of introns) and have GT at the donor site (just after the 5' end of the cut) and AG at the acceptor site (just before the 3' end of the cut). The rarer variants GC/AG and AT/AC are also considered canonical and all other dinucleotide combinations are non-canonical.