I have two groups of patients; for each patient I have an output file (RNA-seq) contains this information

                  readcounts_union  readcounts_intersectionNotEmpt  genelength  R/FPKM_union    R/FPKM_intersectionNotEmpt
ENSG00000000003.10             8                      8               3688     1.06120392730222        1.05722142116966

ENSG00000000419.8             31                     31               4032     3.76132572546529               3.74721015127372

People say


reads that overlap a single feature are included in the count for that feature; reads that overlap more than one feature are not included in the counts


regions that are shared between features are excluded; reads that overlap a remaining region are included in the count for the feature that includes that region; reads that overlap regions from more than one feature are not included in the counts


readcounts_union normalised for gene length and sequencing depth


readcounts_intersectionNotEmpty normalised for gene length and sequencing depth

I want to do differential expression analysis between groups but I don't know which column I should use for DESeq2

Anybody familiar with such a format?

  • 1
    $\begingroup$ I've never seen this format before, what program generated this file? DESeq2 requires raw counts as input (it does it's own normalization internally), so you should avoid using the FPKM values with DESeq2 $\endgroup$ – Tomas Bencomo Mar 1 '20 at 5:54
  • $\begingroup$ Your life would be easier if you simply used standard formats instead of whatever this is. Literature currently seems to prefer pseudoalignment and/or selective alignment (kallisto, salmon) as quantification method, followed by gene level aggregation and diff. analysis. Why bothering with such non-standard things? Get the raw fastq files and start over. salmon-tximport-DESeq2, fully covered with example code in the DESeq2 vignette and tutorial at Bioc. $\endgroup$ – ATpoint Mar 2 '20 at 18:15
  • $\begingroup$ I don't have access to the raw files, I rather have been given only this format and been asked to give them list of differentially expressed genes $\endgroup$ – Exhausted Mar 2 '20 at 20:04

Either of the first two columns are fine to use, though I personally prefer the "intersectionNotEmpty" one.

As an aside, that looks like a custom format file produced from post-processing htseq-count output.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.