In this work we focus on the problem associated with the estimation of gene expression levels using oligonucleotide microarrays. In short, the expression levels are measured using sets of probes that target specific gene. The problem is that each gene has a different nucleotide structure which is used to design specific probes however the sensitivity of the measurement also depends on this sequence. As a result for some genes we may receive higher or lower signal level then it should be just because of the specific probe structure. In this work we focus on the two most important probes features: the sum of the number of G and C nucleotides as well as the occurrence of specific sequence motif in the probe sequence. We show the impact of those features on the measured signal level and propose a method for correcting errors introduced by these features.