Error in example RNA-Seq Example

Question

[lextallan]

Complete novice here so please excuse ignorance and let me know if I need to provide more information.

I'm following the biostars handbook example for RNAseq data (https://www.biostarhandbook.com/rnaseq-align-data.html) and am getting the following error in my terminal:

Error in .local(object, ...) :    NAs found in size factors. Have you called already 'estimateSizeFactors'? Calls: estimateDispersions -> estimateDispersions -> .local Execution halted

when running the command

cat simple_counts.txt | Rscript deseq1.r 3x3 > results.txt

resulting in an empty results.txt file. Thanks!

[h.mon]

There is an issue tracker for this kind of questions. A similar question has been asked before:

DESeq problem!! #103

Please follow the suggestion in that thread, and if doesn't solve your issue, open a new issue there, describing your problem in detail.

[antonioggsousa]

Hi,

Sorry, I don't have access to the book.

Are you just reproducing the tutorial with the tutorial data or are you reproducing the tutorial but with your own data?

António

[lextallan]

I'm using the tutorial data. I ended up redoing everything I could think of in a new terminal window and it seems to have worked now (although I somehow got 290 differentially expressed genes instead of the tutorial's 293).

[antonioggsousa]

The different no. of genes that you're getting it may be related with a different seed. I don't know if the R script that you're using deseq1.r sets or not the computational seed to the same number.

If it does not set the seed to the same number, every time that you run that script you will or you can get a different no. of differentially expressed genes. This is due to some randomness associated with models/methods that are part of the DESeq (that I guess that is the tool that you're using in deseq1.r).

António

[h.mon]

Good you were able to solve your issue - I didn't see your post when I posted my comment. Regarding the differences, it could be due to different versions of the software used in your analysis versus the book.

[lextallan]

It actually fits in nicely with the books' various warnings about reproducibility and variability among tools. Thanks!