Note on null and alternate model:

If one of the Query CDSs overlaps in the same read-frame one of the input 'Known CDS(s)', then don't be surprised if the query CDS gives a negative signal. Such CDSs have already been taken into account as part of the null model. The negative signal supports the null model (i.e. that only the 'Known CDS(s)' are coding) rather than the alternate model (effectively that the Query CDS is coding 'twice').