TOGA multiple codon alignment data

[meilanglang]

Thanks for your work "Integrating gene annotation with orthology inference at scale". I want to use the multiple codon alignment data that use human gene annotation as reference. But when i try to understand the data,something confuse me. the alignment fasta file of a gene seems not have the reference gene codon region sequence. I try to get the reference sequence by transcript ID from Eesembl release 104(it seems the the paper uesd). but the download sequence not equal to the multiple codon alignment file sequence length. for example:transcript ENST00000684760
CDS region i download from Ensembl has 3846nt, but the codon alignment file sequence length is 4570nt. So is the release of Ensemb is wrong or other reason. thank you.
TOGA download link:Index of /download/TOGA/human_hg38_reference/MultipleCodonAlignments
Ensemb:Index of /pub/release-104/fasta/homo_sapiens/cds

[MichaelHiller]

thanks for your question.
All alignments generated with human as the reference do have human.
However, it seems we listed the human sequence as >REFERENCE

Sometimes the reference is not at the top.

I'll add a note late to the README, as this may confuse others as well.
If you want, you can use
sed 's/REFERENCE/vs_hg38/g' -i $file
to replace them all.

I added a note about this in the README.