Statistical Significance for Genome-Wide Experiments

摘要: With the increase in genome-wide experiments and sequencing of multiple genomes, analysis large data sets has become commonplace biology. It is often case that thousands features a set are tested against some null hypothesis, where many expected to be significant. Here we propose an approach statistical significance sets, based on concept false discovery rate. This offers sensible balance between number true findings positives automatically calibrated easily interpreted. In doing so, measure called q-value associated with each feature addition traditional p-value. Our avoids flood positive results, while offering more liberal criterion than what been used genome scans for linkage.