Validation

Can we predict diversity accurately when we know the answer? The strategy for validation of diversity estimators was to sample from distributions having known diversity parameters, including the underlying probability distribution function (pdf). Denoting the pdf as $g(x)$, where $x$ is transcript abundance, we know diversity of transcripts $S=\sum_x{g(x)}$ and population size $N=\sum_x{x \cdot g(x)}$ for discrete distributions having $x \ge 1$.