Protocol Online logo
Top : Forum Archives: : Macroarray and Microarray

validation - struggling to validate illumina mouse ref6!!!! (Dec/18/2008 )

hi all. long time no post for me....
well, brief rundown - I've transfected cells and compared biological duplicates of mock-transfected, neg/scrambled-transfected and test-transfected on an illumina ref6 platform. got a trusty bioinformatician to help out with the analysis. best changes were only 3 fold. applied an arbitrary 1.5 fold (up or down) cutoff. took the genes of interest from the top hits in the mock/neg normalised (ie. only those changed in the presence of my test) and have done some TaqMan based real time PCR. Not validating.
So, do i assume i have entered into splice variant territory? Is a 2-3 fold change too low to confidently expect confirmation?
Where do I go now?!!! Have already wasted 6 samples with an experiment where the mock and test clustered together but not with the neg!!! (I should clarify, that we send our RNA away - so all I am able to do is extract RNA and check it, then get the data back)....
should I push on based on the literature suggested 'top' genes?
should I abort the mission? eeek, that is a conversation I don't want to have with the boss.
I really need to get something - my contract is up in Feb and this was supposed to sort everything out and help to round off my project and set up for the next grant proposal due in Jan!!!!
any advice welcome!!!



well not an expert at all with the illumina ref6 platform or any other array platform - but still... :

So not sure what 'not validating' means in your case but I had quite some differences in LightCyler validations using different primers...; e.g. it could strongly depend on the integrity of you RNA (cDNA), since degradation often occurs from 5' end you might want to try primers sitting more 3'.
2-3 fold hmmm, my experience is that depending on the method the bioinformatician used to analyse the arrays, often they tend to underestimate the 'true' fold change. So, 2-3 fold should be fine.
Since you are using duplicates I am not sure how strong your statistics are for the results.

It can be many things - maybe you can explain a bit more - do you see anything validating in Taqman or nothing etc... etc.. .

Best regards and good luck!