[ensembl-dev] phase of first exon in transcript is 1 or 2

[Jeff Hussmann]

Hello all -

The API documentation for Bio::EnsEMBL::Exon::phase says: 'The Ensembl
phase convention can be thought of as "the number of bases of the
first codon which are on the previous exon".'

Given this, I was surprised to find that a large number of human
transcripts have a first exon with a starting phase of 1 or 2. Why is
this the case? Of course, first exons do not have a previous exon. I
am having trouble imagining experimental evidence that would lead to
frequent annotations of this kind.

Thanks,
Jeff