Scoring function

A score varying between 0 and 1 is assigned to each interaction archived in mentha. The score takes into account all the aggregated experimental evidence retrieved from the different databases. The score is calculated, as defined in the MINT database¹, as a function of the cumulative evidence (x) as:

a determines the initial slope of the curve. We arbitrarily chose a=1.4 in order to obtain a convenient dynamic range of the score distribution.

The x exponent represents the combined experimental evidence and it is obtained by adding up all the experimental evidence weighted by specific coefficients that consider the type of experimental approach and the size of the experiment:

Coefficient definition

• d reflects the size of the experiment. Experiments are defined large scale if the article reporting them reports more than 50 interactions otherwise they are defined small scale. This coefficient is set to 1 for small scale and to 0.5 for large scale experiments.

• e depends on the type of experiment supporting the interaction and emphasizes evidences of direct interaction (e=1) with respect to experimental support that does not provide unequivocal evidence of direct interaction, i.e co-ip, pull down etc (e=0.5).

• x takes into account the number of different publications (n) supporting the interaction.

The MINT scoring function assigns a score close to 1 only to interactions supported by many different reports and experimental approaches while an interaction supported, for instance, by a single high throughput pull down experiment will receive a score of 0.3.

References