Figure 7

Performance on the Latin Square data. Log concentration vs. mean log estimated expression of spike-in genes in the HGU95A (left) and HGU133A (right) Latin Square data sets. There is an increased ability to detect concentration changes at levels lower than 4 pM in the HGU95A data set using GCBGX instead of BGX. Both BGX and GCBGX are more sensitive to changes within the low range than RMA or GCRMA and more sensitive to changes between 0.5 pM and 16 pM than MAS5. The MAS5 concentration line was shifted down by 2.5 units to facilitate the comparison of its gradient with those of the other lines.