When analysis of the library is carried out, the ARCC for the randomly assigned data set should approximate the probability that an isolate would be assigned to a source category by chance.
The ARCC of these removed isolates was then calculated.
The ARCC of the ARA and PFGE libraries was 77.1% and 89.7% with clonal isolates, respectively, and 71.3% and 82.4% without clonal isolates, respectively.
When the ARA library was classified using a 2-way split of human versus nonhuman, the ARCC was 80.2% (68.3% RCC for human and 92.1% RCC for nonhuman).
When the ARA library was analyzed for artificial clustering, the ARCC was 24.8%, only 4.8% greater than random distribution of 20%, indicating that the library was of sufficient size to be used to classify unknown source isolates from water samples.