Ok this explanation is useful, could you suggest a way to plot the pairwise correlation plots?
Hi
supposing I have performed an RNA Seq experiment is there a way to check that the final results (Cuffdiff output ) is correct ?
Supposing after Cuffdiff output I have obtained CummeRbund expression plots: could some one explain what these means (links attached:}
am looking at the CummeRbund Manual but this does not explain into detail the meaning of these expression plots Based on these expression images (links attached) could someone explain what these plots means for the various conditions in respest to the FPKM
Thanks
Beny
1 answer
Box plots of FPKM values shows spread of fpkm values. You can see that between the replicates median values (middle line in box plot) are very close to each other suggesting that your replicates are quite good. Your last two samples conpare to other samples have broad range of fpkm values. But again replicates are beghaving quite similar. You should only worry if box plots of two replicates are very different from each other
Your JS distance heatmap is again similar thing which is measuring how good your replicates are. You can see diagonal line is 0 suggesting identical samples. Lower the value between replicates good indication of replicates are matching.
If you want to check how good your replicates are I would suggest you to plot pair wise correlations between replicates. Generally correlation more than 0.9 is good indicator of good replicates
There are many ways to do that. You can even do pairwise corr in cummeRbund. Other way is use corr function in R which takes FPKM matrix as an input and will give you pairwise corr matrix in output.
Log in to answer this question.