How DESeq2 Handles counts

[TheSeqGeek]

Hypothetically

If I have 24 samples and I submit them to a facility to run my RNA-seq then can this happen?

The technician puts in varying amount of prepped library into the lanes so how does DESeq2 account for this variability in it's normalization?

If I put in 20 ng if DNA in lane one and 50 ng in lane 2 how is this accounted for by the software or is this something I should be aware of and account for by myself.

[dpryan]

The normal packages (DESeq2, edgeR, etc.) will deal with this just fine and it's not something you normally need to concern yourself with. The method used in DESeq2 was described in the original DESeq paper (the end of page 2 and the beginning of page 3).

Having said that, it's often good to have a look at the size factors. It's sometimes the case that if that of one sample differs by 10x or more from the others that you get spurious results.

[TheSeqGeek]

The normal packages (DESeq2, edgeR, etc.) will deal with this just fine and it's not something you normally need to concern yourself with. The method used in DESeq2 was described in the original DESeq paper (the end of page 2 and the beginning of page 3).

Having said that, it's often good to have a look at the size factors. It's sometimes the case that if that of one sample differs by 10x or more from the others that you get spurious results.

thanks you have been a great PI in my research so far


quick code in R for size factors?

[dpryan]

sizeFactors(dds)

where "dds" is a "DESeqDataSet" object.

[TheSeqGeek]

excellent thank you

size factors are around 0.9 - 1.1 bla bla bla