Can we pass custom trained embedding as entity and then train the model?

[MathewKevin]

Hi, we are working on a link prediction task using ampligraph, the triple basically looks like,

protein_sequence, i.e KFLEACD (subject) ---> positive(predicate) ---> assay_1(object)

we already have a better representation of the protein sequence stored as embedding, is it possible to pass those embedding directly as an entity like,

protein_embedding(subject) ---> positive(predicate) ---> assay_1(object)

Please clarify, thanks in advance.