摘要:
145 U133A 2.0 arrays, we used the `hgu133a2frmavecs' 146 R package [42]. All other fRMA parameters were 147 kept as the default options. Probe set expression 148 values were summarized as log2 of the expression 149 intensity. We compiled an aggregate matrix of expres- 150 sion values across datasets using like probe sets 151 and batch effects were removed using the `limma' 152 R package function `removeBatchEffect' where the 153 batches were defined as the 3 datasets [43]. Finally, 154 gene expression values were calculated by aver- 155 aging the probe sets that mapped to the same
展开