DDimerTest

Ddimertest is a statistical method designed to assess evidence for dimerization events in datasets that record pairwise interactions, such as protein-protein interaction networks, chemical association data, or social dyads. The method yields a measure of significance for the presence of dimers and is commonly implemented as a permutation-based test against a null model that preserves marginal distributions while randomizing pairings.

Methodology: The typical workflow starts with a dataset of observed pair interactions and a defined dyadic score that quantifies the strength or frequency of each interaction. A null distribution is generated by repeatedly permuting the labels that designate partner identities, recalculating the dyadic score for each permutation. The p-value is the fraction of permutations whose score equals or exceeds the observed score. The test can accommodate covariates and can be adapted to one- or two-tailed hypotheses.

Applications: Ddimertest is applied in proteomics to identify protein dimers, in computational chemistry to study self-association phenomena, and in network science to detect densely connected dyads. It is valued for its nonparametric nature and its flexibility regarding data structure.

History: The term Ddimertest appeared in illustrative methodological discussions to demonstrate how permutation tests can be used for dyadic association problems. While not tied to a single real-world implementation, several software packages provide similar permutation-based tools for dyadic analysis.

Limitations and considerations: Results depend on the choice of null model and score; the method assumes exchangeability under permutation; sparse data or highly imbalanced dyads can reduce power; computational cost grows with permutation count. Related approaches include bootstrap tests and model-based likelihood ratio tests.