When retrieving all cds sequences from the human reference genome, in the overwhelming majority of cases, the cds is mod 3, starts with ATG, and ends in a stop codon.
However, approximately 10% are not divisible by 3 and have no stop codon (but do have a start codon in the first exon).
Are these likely to be errors?
To retrieve the sequences, I used the Table Browser, i.e.
Select -- refSeq genes for the track and ccdsInfo for the table Under output, select "selected fields from primary…" then click get output You will go to another page that gives you the option to select additional tables -- Select "ccdsGene" -- click "Allow Selection from…" That will expand another list of fields...etc.
-- click "Check all" at the top click "Get output"