Real-valued HTseq count to DESeq2

Question

[english.server]

HI!

In order to input non-integer values to DESEQ2, the thread at

https://support.bioconductor.org/p/105964/

suggests that one rounds the values. I am not much familiar with RNA-seq analysis pipelines but I suppose HT-seq count should give integers as row counts or as I saw somewhere at biostars.org real values as count estimates. I am not sure whether it is possible to round data available at

https://xenabrowser.net/datapages/?dataset=TCGA-UCS%2FXena_Matrices%2FTCGA-UCS.htseq_counts.tsv&host=https%3A%2F%2Fgdc.xenahubs.net&removeHub=https%3A%2F%2Fxena.treehouse.gi.ucsc.edu%3A443

and then pass it to DESEQ2, ie is the thread at bioconductor website also applicable to this sort of data? The dataset is gene expression RNAseq - HTSeq - Counts from TCGA Uterine Carcinosarcoma (UCS).

Thank you

Answer 1

[Kevin Blighe]

I presume that it was you who just up-voted my other [related] thread this morning: A: Normalisation of RNAseq data from UCSC Xena Browser ?

In that thread, I also recommend to round the counts to 0 decimal places, i.e., round them to integer counts.

The Xena data is not integer counts because they [UCSC Xena] increment (+1) the raw HTSeq counts from Genomic Data Commons (GDC), and then log [base 2] transform these.

It is perfectly fine to revert these back to their original form, and then round to 0 decimal places.

Kevin

[english.server]

Yes that was me. Thank you for your time and confirming on the matter.