Recently we performed miRNA microarray and we found that many novel miRNA candidates are present in our data. Although many of them have high expression value, their functions are still undefined and we don't know whether they exist. Should we consider them during normalization?
If you are using the whole of the miRNAs as reference, then: yes. The idea is that *on average* the level of the miRNAs should not change much. If several hundered miRNAs a re measures (and used for normalization), then the regulation of some of them won't be influential (especially when some might be induced and others might be repressed). However, if you assume that there is a considerable "net-induction" (or repression) or the majority of the analyzed miRNAs, then you would need something else to normalize against (anyway).
Usually U6 small RNA is used as internal normalized control.
For example, U6 sRNA is in Agilent miRNA microarray.
You may also try to perform stem-loop RT-PCR to confirm some of miRNAs' expression level (miR-103 as control).
This paper may also help: Normalization of microRNA expression levels in
quantitative RT-PCR assays: Identification of suitable reference RNA targets in normal and cancerous human solid tissues. [RNA (2008), 14:844–852.]
Thank you very much Professors for your valuable comments. In my data, total signal intensity in treated vs control does not change much. So, after background subtraction, we performed global median normalization and then checked the level of some miRNAs by quantitative stem-loop RT PCR. Except some highly upregulated/downregulated miRNAs, many of them don`t have consistency with array data. That made me feel whether those novel miRNA candidates with high expression value have any impact. May be I should check other methods.
Hi Ruhul,
I think the data normalization is only a part of the processing that you want to perform before you start analysing your data. If you can follow a simple workflow outlined below you should have no doubt about your findings. I have analysed more than 50 pilot/large scale studies with up to 3000 arrays. If you have some repeat samples in your study that would be great because they will tell you which normalization method makes your data better. E.g. repeat samples should give you the same reading (which they never do) and now you start to apply different normalization techniques and work out which one works best.
My general approach to data normalization is part of the data exploration which I apply before moving on to data analysis.
1) Data exploration (sometime called data clean-up)
2) Data transformation
3) Outlier detection and removal
4) Data normalization
5) Data analysis
6) Validation of results
The initial step in data transformation could be a simple analysis along the line of generating set of random numbers that belongs to a certain distribution family (generate the same number of data points as in your data set) and then plot it in a qq-plot. It will be very easy to spot which distribution family your dataset belongs to and then you can transform your data.
I really like to plot the normal transformed data in a histogram and it can be very helpful, because I came across a number of platform that contain systematic technical issues. A histogram of the normal transformed data is one way of spotting them. Often this technical issues show as humps or sharp peaks (zero readings) or on the left hand side of the histogram as a max reading value on the right.
Other ways for the identification of outliers could be PCA or boxplots….another of getting ride off your right hand tail is to apply a data normalization procedure.
After this you should be able to analysis your data. If not feel free to get in touch and we can discuss it offline.
Best wishes,
Jens
You can find some more tips and tricks on the webpage below.
http://www.biomarker-discovery.com/data-analysis-overview
www.biomarker-discovery.com
- Badges
- Science topic