Micoarray meta data analysis

Question

[au.rinki.bio]

hello, I am working on a meta-analysis of the microarray data for the different type of cancer and trying to find out common genes for these cancers. For the same type of cancer (high grade) from different data sets, we get a different number of genes, as we increase the number of data sets, the common genes among these data sets were reduced. why this is occurring? If the different experiment gives the different results. how much the accuracy of these experiments?

[WouterDeCoster]

Don't you expect different results as you are working on multiple cancer types?

[au.rinki.bio]

here, I am asking for same cancer (cervical) but datasets are different.

[Kevin Blighe]

Reproducibility is a constant issue in biomedical research. You should never expect that 2 studies looking at the same cancer sub-type will produce the same results. There are so many factors going into these experiments, many of which will never be reported - these factors are responsible for biases.

When you are analysing each dataset independently, you can then perform a 'meta analysis' of the results using RankProd.

Answer 1

[guangyan.zh0]

You should try gene-set level meta-analysis, it will be less affected by the small number of common genes shared across datasets.