Hello everyone,
I'm currently studying a certain synapse and, apart from the functional characterization, I want to show that a receptor is present in a certain type of cell.
Recently, 4 different papers have been published that perform single-cell RNAseq in that kind of cells with slightly different controls and all of them have made available both the raw data and the pre-processed data with the normalized and corrected relative expression levels.
As my only intention is to show that 2 genes are being diferentially expressed in my cells of interest, I thought about showing the log 2 fold-change in expression for my genes of interest vs. the corresponding control for each study.
Is this acceptable or I have to inevitably perform the analysis from scratch with the raw data from each study?
thanks in advance!
Hi Augustin
if the analysis from the papers you cited has well been done, yes you can use those data to perform your own analysis as it is done in GEO profile ou GTex, but at anytime you can mix those advanced data since protocols and biases are not same. a meta-analysis should be done.
fred
In general, omics data needs to normalize properly due technical noises. If you want to merge a number of different datasets from different studies, only normalization will not be enough. You need to minimize the batch effects. Many meta-analysis without batch effect minimizing are getting publish though. They are full of biases. Applying log2 for normalization might not be enough. Check the normality of data. If not normal/semi-normal, you need to use rule-base normalization in addition. After that, you can minimize batch effect using comBat. See batch-effect part here ( Article Unbiased data analytic strategies to improve biomarker disco...
)- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication