Determining whether a measurement signature is specific to a biological process

A Specificity statistic (or metric) is computed as a means to identify amplitude scores associated with a signature that can be attributed with high probability to a specific biological entity or process represented by the signature. Preferably, Specificity is computed by assessing a likelihood of a given null hypothesis, namely, that an amplitude score is not representative of the specific signature but, instead, is representative of a general trend in the applicable data set that can be measured by any signature that is comparable to the signature of interest. In a typical implementation, a first step to compute the Specificity metric is to construct a set of comparable signatures. Next, an amplitude score is computed for each of these signatures, preferably using the same data set. Then, the Specificity metric is computed, preferably as a two-tailed p-value, by placing the amplitude score for the signature of interest on a distribution of scores for the comparable signatures. Scores that have Specificity p-values less than a particular value, e.g., 0.05, are considered to be scores that can be attributed with high confidence to the signature of interest.