Missing sample in Cuffdiff time series

[colind]

Hello everybody,

I am new to the RNAseq and trying to use the tophat-cufflinks pipeline.

Now I have the RNAseq data from 5 animals on Day -2, 3, 5, and 8. I am thinking about using cuffdiff to do a time-series analysis. But I missed one sample: I have only 4 samples from Day 8. The animal died before Day 8.

I am not sure how to do the time-series analysis in this case. Do I have to keep 4 replicates for each day, or I can still run all the samples I have even one sample is missing?

Please give me some suggestions. Thank you every body!

[mbblack]

Hello everybody,

I am new to the RNAseq and trying to use the tophat-cufflinks pipeline.

Now I have the RNAseq data from 5 animals on Day -2, 3, 5, and 8. I am thinking about using cuffdiff to do a time-series analysis. But I missed one sample: I have only 4 samples from Day 8. The animal died before Day 8.

I am not sure how to do the time-series analysis in this case. Do I have to keep 4 replicates for each day, or I can still run all the samples I have even one sample is missing?

Please give me some suggestions. Thank you every body!

You can run cuffdiff with unbalanced replicates. Its a perfectly legitimate analysis. With a minimum of four in your data set, you are still on reasonable, solid statistical ground for your comparisons. Significance for each condition will be based on an independent average for each condition (as explained in the "howitworks" section of the FAQ) so it will account for the differences in "n" and "m" in each condition.

[colind]

thank you very much!