For extreme datasets, containing either redundant or complementary features as well as noise, the index value is defined by mathematical formula.
In particular, increasing the number of imputations increases the chance of drawing an extreme imputed dataset that is assigned an extreme weight.
In general, it seems that as we increase the number of imputations, the more likely we are to draw a really extreme imputed dataset which is assigned nearly all the weight.
It was also performed between two groups of two extreme fetuses (dataset 2 excluding the DM108 heterozygote and an NM0423 animal, the myostatin expression and gene profiling of which were intermediate) to maximize the difference between the two genotypes (Table 2).
In the extreme case, datasets (and specified applicability domains) for two models can be disjoint.
The method also appears to be very robust to more extreme simulated datasets, with 100% successful estimation.
At the opposite extreme, large datasets support a clear decision about the topology of the tree, placing most of the mass of the posterior distribution in a single orthant, and resulting in mostly agreeing estimates.
At the other extreme the entire dataset is made available to the scientific community.
In the absence of such a dataset, extreme phenotype study design, in which samples from both ends of the phenotype distribution are analyzed, can serve as a suitable alternative [ 93].
For example, ppp-values close to zero indicate that most of the replicated datasets were less extreme than the observed datasets, and hence the model shows less discrepancy from predicted values, compared to the real data.
