However, with PSEA the larger the variability in the abundance of a particular population, the greater the sensitivity of specific expression detection in this population.
A posteriori, we checked for the systematic influence of any single sample on the results of PSEA by inspecting the distributions of Cook's distance [ 39] for all genes whose expression could be successfully deconvoluted (see below).
In the case of the putative PolII transcripts, the high-scoring PSEA elements preferentially locate 50 nts upstream of the transcript start.
In case of U4atac, the PSEB and PSEA segments were spatially inverted, i.e. PSEB was located upstream of PSEA.
Furthermore, the noise in marker gene expression measure decreased the sensitivity of PSEA (see Methods).
Therefore, they allowed us to test the applicability of PSEA to a standard gene expression dataset from a human tissue-based expression study.
For DTLZ test suites, the values of the H indicator obtained by PSEA and PSEA-m with respect to the number of generations have the same tendency of changes.
Together with the fact that overall PSEA more efficiently detected contribution from the more abundant population (i.e. granule cells, see above), we concluded that expression contribution by cell populations of lower abundance could effectively be masked by expression in a more abundant cell population.
We compare the values of the H indicator obtained by PSEA and PSEA-m with respect to the number of generations in this subsection.
The second one, called PSEA and proposed in [27], carries out set-based selection and crossover guided by preferred regions as well as polynomial mutation.
