Can DESEQ tool take in normalized read count (real value number) instead of raw counts (integer types)? I don't want to round my normalized table to the nearest integer because I don't want to lose any information so i was wondering if there is a way to input a table of normalized read counts?
thanks,
DESeq uses the countDataSet class to store the count data and the sizefactors plus other meta data. You can understand how to create a countDataSet from here under countDataSet class.
However, I am not sure about the normalization method you used and if by any means that you can create a class of this type with appropriate metadata, it would be possible to use the DESeq package for further analysis.
However, for normalization, the DESeq computes the sizeFactors and in the same document you can see that it is possible to manually specify the size factors (under the function estimateSizeFactors), but nothing much is dealt in that regard and it is usually not a case that one would come across quite often.
temp = t(sizeFactors(cds))
sizematrix<-matrix(data=temp, nrow=nrow(RawCountData), ncol=ncol(temp), byrow=TRUE)
scaledCounts = RawCountData/sizematrix
You may also want to compare the normalized counts that you have with that of the DESeq for your raw counts to see the impact. Above all, your concern towards loosing significance due to decimal point is something I don't understand.
You might want to give more information about exactly how you normalized things. If you did, for example, conditional quantile normalization, then use DESeq2 instead of DESeq and see the end of the DESeq2 vignette for a usage example.
Otherwise and as a general rule, don't try to cram non-integer values (e.g. rpkm/fpkm) into a statistical model that relies on integer values for its validity.
No, you cannot input transmogrified numbers into methods that assume count data
You must input raw count data into DESeq and edgeR.
If you have some normalized or otherwise transmogrified version of your raw data that you insist on using, then your best bet might be to use limma::voom.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
What if the read counts were normalized using DESeq? Is there a way to start form the already normalized counts? A colleague did an initial DESeq2 run on our data, and supplied a table with the normalized read counts from DESEq. I do not have the raw counts, and was wondering if I could start from here and skip the normalization method that DESeq perfoms.