Could you normalize data from microarrays to the degree that it would be possible to compare the different sample values with sample values from other studies?
Say you want to find what gene expression different cancers shares, so experiment/study 1 is comparing brain cancer and liver cancer, so sample one comes from a brain tumor and sample 2 from a liver tumor, in the second experiment/study, which is unrelated to the first one, lung cancer and skin cancer is compared.
So I want to cross-compare all the different samples, not just sample 1 with sample 2 from experiment/study 1, but sample 1 from experiment/study 1 with sample 1 and 2 in experiment/study 2 as well as sample 2 in experiment/study 1 with sample 1 and 2 in experiment/study 2.
I was thinking about finding a reference gene, a house keeping gene that is expressed equally much in all tissues and cell types but this proved difficult. So maybe you could find an average over all the different experiments and then multiply or divide the values in each sample (I know this is a longshot..), depending on if they are below or above average, or in any other way make them comparable to each other. Ideas?
Would this be possible with affymetrix one color maybe?
I looked in to microarray meta analysis, but this is somewhat different since meta analysis focuses on pretty much just combining the data from two different experiments to make them statistically stronger, that is, adding the data from sample 1 from experiment/study 1 with sample 1 from experiment/study 2 and sample 2 from experiment/study 1 with sample 2 from experiment/study 2, in studies that has a reference or control, sort of. From what I understand..
I am novice to microarrays and meta analysis as well as statistics, so please respond plainly and feel free to correct me if I made false asumptions and redirect me if this information is to be found somewhere else!