Entering edit mode
11.0 years ago
DevS
•
0
I want to know how relevant is it to take control samples when I wish to find only the differentially expressed genes among 3 or more types of cancer. When I go for further analysis like the pathways involved in those cancer from the DE genes or TFs involved, then will not taking control affect it? And what should be the approach to validate some genes from those (as there is no control) in any one of the 3 or more cancers? I am new to microarray..pls help.
Thank you Irsan. Then how do I validate some of those genes? Say I get "n" number of genes DE between three cancer type1, type 2 and type 3. How do i go for validation? qRT-PCR? Against what should I check the expression?
What do you want to validate? Do you want to validate that the differences yielded by micro-array are no technical artifacts or do you want to validate whether the microarray observed differences are relevant fir tumor progression?
I want to validate that the differences yielded by micro-array are no technical artifacts. I mean if I get that a gene/a cluster of genes is/are over expressed (high intensity values) for a particular cancer type, then validate that its true. Can you pls also tell me how validate whether the micro-array observed differences are relevant for tumor progression?
For the validation you can do a qpcr between tumor a and b and do a two sample two tailed t-test (assuming qpcr results are normally distributed which might be tricky, maybe a non-parametric test would be more appropriate). If you have more than two tumor types use one-way anova. However when qPCR is not significant it does not mean that the micro array is unright because the qPCR test does not have 100% statistical power. If qPCR does turn out to be positive you are more sure about your micro-array (use false discovery rate adjustments for micro-arrays!). If you want to judge whether observed differences contribute to tumor progression you have 2 options: 1) look at the functions of the genes that are different, are many of them related to cancer pathways in the past? Use ontology enrichment analyses with e.g. DAVID. 2) include the proper controls in your experiment, in this case tumor (sub)type matched non-malignant cells
Thank you so much for the detailed explanation. :) I used to read in papers about the statistical analysis after qPCR, but was never able to get why it was needed! And yes for seeing if the genes are involved in tumor progression I have done functional enrichment analysis using IPA. I was not sure how to validate it... One more thing- say that I have found from microarray that gene 1 (gene related to cancer) has a high intensity expression (which is considered to be proportional to the amount of mRNA present) in cancer a . Now when I use proper control for the cancer a and go for qPCR to validate the gene, then am I also supposed to get that the gene is up-regulated when compared in control? ( which i think should be given that its a cancer related gene and seen to be over expressed in the cancer..still pls say what should I expect). Thank you
Yes, that could be the case. In my experience however most of the big differences you find between different tumor types are differences that reflect the cell type of origin/tissue.
Hey thank you so much! :)