With merge request !79 (merged), a bug was introduced in the cds building process (which is used for the translation to protein). All annotations that were added with the version since !79 (merged) and this version will have wrong protein translations for genes on the negative strand with multiple exons. This merge request fixes the code for them.
To be precise, the following two bugs are fixed here:
- The reverse complement was calculated but never used.
- The new CDS sequences were always added at the end of the cdsBuilder, whereas they should be put in front when looking at the negative strand.
Furthermore, we discovered that
pantools add_annotations still crashes with trans-spliced genes. This appeared to be caused by trying to delete a nucleotide node. This will also be fixed here.