While I also think that GO term prediction is useful/interesting to many users, the boxes can become very large depending on the annotation depth of the protein we copy annotations from.
For a more well-defined user experience, we might want to make the prediction of subcell. default again (this will always have the same size and it comes with a nice visualization).
To make the problem about very large result tables for GO prediction more explicit, check GO predictions in EMPP for https://www.uniprot.org/uniprotkb/Q9NZC2/entry
While I also think that GO term prediction is useful/interesting to many users, the boxes can become very large depending on the annotation depth of the protein we copy annotations from. For a more well-defined user experience, we might want to make the prediction of subcell. default again (this will always have the same size and it comes with a nice visualization). To make the problem about very large result tables for GO prediction more explicit, check GO predictions in EMPP for https://www.uniprot.org/uniprotkb/Q9NZC2/entry