I have found a significant gene (BIRC5) between lung cancer tumor and adjacent normal samples. In the literature, it is mentioned that BIRC5 (an anti-apopotic gene) is generally upregulatd is most solid tumors. But unlikely, in my case I'm getting its expression as downregulated in tumor samples as compared to normal. Is it possible to get expression status of a gene altered in biofomatics analysis as compared to wet-lab results? Does this depend on sample size or type of dataset?