I ran my annotation using ESTs from NCBI and transcriptome data and Swiss-prot data. I used SNAP trained twice after which Augustus was trained once.
I tried to submit the annotated genome to NCBI but have some problems I have not experienced before and do not know how to solve.
Some of the CDS features have invalid translations. Every CDS feature should have a valid start and stop codon, and should not have any internal stops. The only exception is if the CDS is partial at the end of a sequence or at an intron/exon boundary. If the CDS is partial, it must have the appropriate partial symbols.
If a CDS feature does not have a valid translation (for example if there is a frameshift), please remove the CDS feature and annotate this with a single gene feature across the entire span. Include a note on the gene with a brief description. For example:
1 200 gene gene phoA gene_desc alkaline phosphatase locus_tag OBB_0001 note nonfunctional due to frameshift
 There are 1075 gene features that are not associated with any other features (CDS, rRNA, etc.) and are not labelled as pseudogenes or as nonfunctional genes. Did you lose some of the annotation you intended to include.
Please how can this be solved?