Direction of Interactions (Bait/Hit)

Bait-Hit Curation

When curating the interactions, we are applying a “spoke” model that directly pairs bait proteins with associated proteins (like the spokes of a wheel). Using this bait-hit approach, we are also providing some directionality to the interactions. The bait can be entered in the 'bait' column of the tool, and any relevant interacting protein(s) can be listed in the 'hit' column separated by a return. There are different options for the bait-hit entries: one to one separated by rows (the number of baits must equal the number of hits), one bait to many hits, or else many hits to one bait.

For some experiments, e.g. two-hybrid studies, the bait and hit are obvious. In this case, the bait protein is fused to a DNA-binding domain (DNA-BD) and should be entered in the first column (bait). The hit/prey is a library fused to a DNA-activation domain (DNA-AD), and any interacting proteins should be entered in the second column (hit/s) on separate lines. Similarly, if a paper shows an in vitro kinase assay, the kinase would be entered as the bait, and the subtrate as the hit. Note that we curate reciprocal interactions if the bait-prey directionality is clear and also self-self interactions (e.g. dimers).

However, for many interactions, the bait-hit directionality is not clear, e.g. when a paper shows synthetic lethality without doing a specific screen to look for gene B that is synthetic lethal with gene A. For those types of interactions that do not have an inherent directionality, the curator must choose what they think would serve as the best bait (e.g. the gene/protein that is the main focus of the paper), or in some cases just choose a bait at random. In these cases it is critical that the curator only enter the interaction once (with their chosen directionality) and not enter the interaction again with the opposite directionality to prevent misleading redundancy. For genetic interactions where a direction is implied, though, we need to be consistent in the way the interaction is entered. For synthetic/dosage rescue experiments, the 'rescued gene' should be entered as the bait and the 'rescuer' gene as the hit. For example, overexpression of gene A rescues the phenotype of a mutation in gene B. In this case, gene B is the bait and gene A is the hit.

Co-purification experiments may or may not have an obvious bait-hit direction. If there is an actual bait used for the co-purification, e.g. a tagged protein is isolated using affinity purification and any associated proteins also co-purify with this protein during subsequent fractionation steps, then this can be entered in the tool using the tagged protein as the bait and any associated proteins as hits. An example of a purified complex can be found in Fig. 1 of pmid 10777588 in which the bait protein is considered to be Elp2p. Since we are curating the interactions using the spoke model representation, there would be two bait-hit pairs for this co-purification experiment: ELP2-ELP1 and ELP2-ELP3. We are not using a matrix model representation which for this complex would include a third interaction: ELP2-ELP1, ELP2-ELP3, and ELP1-ELP3. The reason the spoke model is preferable is because complexes with many components will artificially increase the number of interactions in the database if they are entered as matrix representations.

It is only possible to enter the gene name in the tool if the name is unique. If the name is not unique, then the tool will indicate that there is some ambiguity to the name in which case the ORF name will have to be entered. In some cases it might be necessary to refer to a paper that is referenced by the paper you are curating in order to verify the experimental system and also the genes/proteins involved. For example, Fig. 1 of pmid 7479899 references co-purification methods that are displayed in Fig. 1 and table 1 of pmid 8187178. Going forward, we're also going to curate cross-species interactions, e.g. between a yeast and a human protein. The curation tool allows us to select separate species for the bait and hit using the relevant pulldown menus. However, this does not include cross-species complementation experiments because they are not really genetic interactions between two genes, but rather a test to determine functional orthologs in other species.


It is also important to note that if there is no obvious bait, for example in a co-fractionation (CF), co-purification (CP), reconstituted complex (RC), or co-crystallization(CC) experiment, then we can enter the information as a complex. This can be done by listing the members of the complex in the bait column and selecting the 'Add as Complex' option at the bottom of the form (e.g., see the 'Active Complexes' entry for pmid 18971323). This option only applies to protein complexes. If the interaction involves 3 or more proteins, then we can enter these as complexes, but interactions involving only 2 proteins should be entered in the standard manner, regardless of the experimental system (reconstituted complex, co-crystal structure, etc.) and the bait should be chosen by the curator.

Back to BioGRID Curation Guide

curation_guide/direction_of_interactions.txt · Last modified: 2017/08/08 12:52 (external edit)