*edit: new post opened here, :) . Feel free to answer there, if you like.

Thank you for the hint!

I downloaded the current BIOGRID-ORGANISM-Saccharomyces_cerevisiae_S288c-3.4.132.mitab and, I see that:

1. there are other methods to detect PPIs, such as: FRET, PCA, Co-fractionation, Co-purification [...];
2. there are entries with no confidence-score provided ("-" at column 15) .

My question is:

Given (2), what methods have been proved to be, among the others, more reliable in detecting PPIs?

Maybe this question is better suited for a separate post.

The listed methods typically considered to show direct interactions are: two-hybrid, reconstituted complex, protein-peptide, PCA, FRET, far Western and biochemical activity. I normally would consider co-crystal structure as indicative of direct interaction, except that here they also include NMR and EM structures which may not have the resolution to resolve direct interactions.
For any interaction, you can retrieve protein sequences using the IDs provided. However, the tab-delimited files only give gene IDs of interacting partners. For protein IDs, you'll need to get the PSI-MI 2.5 files.