作者: Or Zuk , Eytan Domany , Liat Ein-Dor
DOI:
关键词:
摘要: Ranking objects is a simple and natural procedure for organizing data. It often performed by assigning quality score to each object according its relevance the problem at hand. widely used selection, when resources are limited it necessary select subset of most relevant further processing. In real world situations, object's scores calculated from noisy measurements, casting doubt on ranking reliability. We introduce an analytical method assessing influence noise levels use two similarity measures reliability evaluation, Top-K-List overlap Kendall's tau measure, show that former much more sensitive than latter. apply our gene selection in series microarray experiments several cancer types. The results indicate lists obtained these very poor, experiment sizes which attaining reasonably stable Top-K-Lists larger those currently available. Simulations support results.