Hi all,
Is there any possibility that the RNA-seq data do not agree with the qRT-PCR results? I don't have extensive experience in RNA-seq data analysis, thus recently I sent my RNA sample to my collaborator and they did the RNA-seq and analysis work and back to me the processed value. Among the genes, one of the gene should be much lower in one sample vs. the another sample, since I siRNA knocked down the gene specifically in the sample, and my qRT-PCR confirmed it has been knocked down about 70%. But, in the RNA-seq data, the gene levels are almost the same between these two samples! I also checked other genes, that I know will change upon this gene, and their levels look having the right trend in these two samples. Who can explain to me is it possible? They used both Cuffdiff and Gfold methods to analyze.
Thank you!
Xiaoyong
I would not suggest to go for Cuffdiff.
I think if you just look at FPKM levels (Cufflinks output) then you should notice a difference.
Sorry, can you explain more why not go for Cuffdiff, while Cuflinks can show difference. I am not very familiar to these tools. Thanks!
Cuffdiff is fold change calculation
Cufflinks would give you the normalized expression value (FPKM), which is based on algorithm which calculates that how many reads mapped per KB per million of reads..
for foldchange I never use Cuffdiff, but I go for DE-Seq algorithms in similar case as yours.
Just come across me mind. Is that possible that RNA-seq also covered the siRNA sequence I used for knockdown in one of the sample, which false-positively increases the reads of the specific gene, that is supposed to be lower than the control knockdown sample?!
Not if its PolyA RNA-seq
That is very good point! I will double check my collaborator about the method they used in the RNA-seq! Thanks!
Even still, you'd only be getting reads targeting the siRNA, not the whole transcript of the target gene.
How many replicates did you run per condition? Did you use the same RNA as input for your RNA-Seq and your PCR?
RNA-seq only has one sample for each condition... I don't have the information of the input RNA in RNA-seq reads, but my qPCR primers are designed to cover ~150 bp, should not recover the siRNA seq. That may explain my qPCR always confirmed the KD efficiency well, but for the RNA-seq...may even cover the small siRNA? I used the RNaeasy kit to extract total RNA, should the small siRNAs (~23bp) not enriched through this extraction. I am very not sure...
Let me rephrase: did the RNA for the RNA-Seq and the RNA for the qPCR come from the same instance of your transfection? E.g. you transfected a flask of cells, hit them with trizol, and sent half for RNA-Seq and kept half for qPCR?
If the RNA-Seq input material came from one run, and the qPCR material came from a second run, there's always the chance that one siRNA knockdown failed while one worked. The RNA-Seq and the qPCR should be run on the same RNA, and the siRNA should be run by the same people, same cell passage and so on. The extractions should be performed by the same person using the same method.
A silly question, but if you used different input material, were the cells harvested at the same time and in the same way? It is possible that 1 replicate may not provide enough statistical power to resolve differences in your gene of interest. Not to mention, n=1 will really reduce your discovery potential, adding replicates increases statistical power and in turn will increase the likelihood that you will correctly identify DE genes. Furthermore, replicates will provide a more robust experiment, if a siRNA knockdown in a given replicate fails, you'll have n-1 samples left.
RNA-Seq is much less fickle than microarrays, but stuff like this can still cause problems. Sequencing has gotten cheaper but it is still expensive, I suggest you work closely with the people running the sequencing to get all of the details and design the best possible experiment.