Yes at times the genbank files can be problematic in that they are incomplete or that BioPerl gets cranky, you have not provided a code that I can test but if you may consider the following workaround, work with the fasta files in conjunction with the feature table provided in the genbank files

• convert the genbank to gff through (genbank2gff3.pl)
• convert the genbank files to fasta or download the fasta equivalent
• parse the gff files and extract the CDs with their coordinate information

  perl -F'\t' -lane 'if($F[2] eq "CDS"){print}' GCA_000153565.1_ASM15356v1_genomic.gbff.gff | cut -f3,4,5,7 > GCA_000153565.1_ASM15356v1_genomic.coordinates.txt

• extract the subsequences from the fasta files using the coordinates saved in GCA_000153565.1_ASM15356v1_genomic.coordinates.txt

For the last item you may use BioPerl::SeqIO $seq->subseq(start..stop) but make sure you get the reverse translation of the seqs in the negative strand