The table on https://github.com/MI2CAST/MI2CAST/blob/master/examples/PKA_regulation.md is an excellent overview of the contents of a MI2CAST compliant file - it's obvious what the entities are and their labels.

Inside the tab file, however, it's no longer obvious the separation between entities structured identifiers (written as CURIEs) and the labels such as in "psi-mi:""MI:0362""(inference)".

Why not split this to two columns, one for MI:0362, and one for inference? For this specific one, the psi-mi: prefix is sort of redundant, since it's well accepted that the MI: prefix is used for this ontology.