Hello,
I'm studying about detection of differential expressed genes (DEGs) by using disease vs healthy samples microarray data. I use Limma in Bioconductor for analyze the DEGs. I realize that some of DEGs are both up and down regulated. For example, while ARAP2 gene was upregulated in 2 probe set, this gene down regulated in 3 probe set at one dataset. How is this situation occur in transcriptome level. Is this gene up or else down regulated in real? How are we explain both up and down regulated genes in same dataset?
Check the sequences of the different probes. It may be that the probes
- do not really target the same RNA
- traget different transcript variants
- show different degrees of cross-hybridization (with other sequences)
- show differing degrees of mRNA degradation (correlation of probe position + oligo-dT priming + mRNA degradation)
This list is surely not comprehensive, but it lists the most obvious possibilities.
There are evaluation functions cited in the litteratrure (look https://www.researchgate.net/publication/268979170_A_NEW_SURVEY_ON_BICLUSTERING_OF_MICROARRAY_DATA?ev=prf_pub )
Besides, you can use available packages on R
or look for GO ...
Conference Paper A New Survey on Biclustering of MicroArray Data
Hi Esra,
I would suggest you not to completely depend on the Microarray data expression pattern. The genes of interest should always be validated by real-time PCR.
Thanks Prof. Wilhelm, maybe, I think that gene expression ratio may change depend on disease progression. Could you also think there is possible?
Tnaks Prof Yuang, I collect data from gene expression omnibus, there are almost 400 sample and 8 different dataset. I detect both up and down regulated genes at all datasets, separately.
Thanks prof. Saber and Prof. Saini. I consider your purposes.
As Ajay told us before, the better way to get a clue is to validate your interest genes by QRT-PCR. Also a very important thing is to validate your experiment with other samples than the ones used in the microarray experiment, otherwise a referee could tell you that you are validating the technic but not your experiment. Hope it helps.
Thanks you are right, validation is required. But I wonder that one gene is both up and down regulated in a dataset (sample). I think, this situaition may be deal with disease mechanisms. Maybe while this gene up regulate , conversly same gene down regulate in disease progression. Could you also think there is possible? I ask this question because this situaution is possible in real Biology.
In my opinion, itś true that one gene could in diferent states of the disease behave in a different way but if you find different replicates of the same state with different fold directions it will probably as Jochen Wilhem said, due to differences with the probes.
Also, you should look not individual replicates but take into account say the average/median of the expression value of all the replicates for the same condition. Having information regarding the dispersion of the data could also make you an idea. It could be that one replicates is behaving as an outlier...
Hi Ersa,
What you have pointed might be right....it is possible in real biology.....but at the end all possibilities needs to be supported by experimental evidence and statistics.
As your a data is from several microarray data sets in the Omnibus database there might be other factors responsible for such observations. Whatever may be the case it needs further careful confirmation. The up and down-regulation of genes might vary during disease progression...or it might be routine metabolic fluctuations affected by many other factors..as you may agree that the genes/gene products do not work in isolation and they are in a complex metabolic network. Here, one need to identify the allowed fluctuations from the linkage of up-/down regulation with the disease and that too with a statistical significance...Hence even if your assumption is correct you need to validate it with additional experiments including qPCR analysis which will be more precise in assessing the precise extent of up-/down regulation of the genes of interest
best regards
Just a nitpicking comment on Ajay's post:
qPCR is not more *precise* than arrays. The precision is rather bad. But the dynamic range is far better, allowing a more *accurate* determination of larger expression changes.
The best (but not always achievable) validation, however, is an experiment showing the biological relevance of the estimated change ;)
Thanks for your answers. All of them are valuable information for me.
- Badges
- Science topic
- Similar topics
- Chemistry
- Biochemistry
- Biochemical Techniques
- Blotting Techniques
- Immunoblotting
- Dyes