question about CDS phase in gff3 output

[Monica Poelchau]

Hi there EVM group,

I have a question about the gff3 output from the EVM program - specifically about the CDS phase.

Consider the following example of a simple mRNA with 2 CDS features. Here's some information about the first CDS feature (assuming + strand, 0-based coordinates):

phase 0

start 0

end 9

CDS sequence length 9

aa sequence length 3

If we had no other information about the next CDS feature in this example mRNA model, then we'd assume that the next CDS feature's phase would be 0. Can we assume this to be the case with EVM output, or could other evidence alter the phase of the next CDS feature in the model to be 1 or 2, instead?

Thanks!

Monica