To detect changes in gene expression data from microarrays, a fixed threshold for fold difference is used widely. However, it is notalways guaranteed that a threshold value which is appropriate for highly expressed genes is suitable for lowly expressed genes. Inthis study, aiming at detecting truly differentially expressed genes from a wide expression range, we proposed an adaptivethreshold method (AT). The adaptive thresholds, which have different values for different expression levels, are calculated basedon two measurements under the same condition. The sensitivity, specificity and false discovery rate (FDR) of AT were investigatedby simulations. The sensitivity and specificity under various noise conditions were greater than 89.7% and 99.32%, respectively.The FDR was smaller than 0.27. These results demonstrated the reliability of the method.
展开▼