DESeq2 design

Question

[Rajesh]

Hi Experts, i have this feature counts matrix for 23 samples (each containing multiple replicates ranging 3-16). Among these 23 samples some are untreated for some to which i want to compare with or call DESeqDataSet. the samples are something as-

run treatment_concen    cell line   conditions  time
SRRxxxx1    0   A1  untreated   24hr
SRRxxxx2    2   A1  treated 24hr

SRRxxxx3    0   A1  untreated   24hr
SRRxxxx4    0.2 A1  treated 24hr

SRRxxxx5    0   A1  untreated   24hr
SRRxxxx6    15  A1  treated 24hr

SRRxxxx7        A1  untreated   9hr
SRRxxxx8        A1  treated 9hr

SRRxxxx9    0   A1  untreated   24hr
SRRxxxx10   2   A1  treated 24hr

SRRxxxx11       H1  healthy 
SRRxxxx12       H1  paitent 

SRRxxxx13   0   A1  untreated   24hr
SRRxxxx14   0.02    A1  treated 24hr

SRRxxxx15   0   A1  untreated   24hr
SRRxxxx16   2   A1  treated 24hr
SRRxxxx17   2   A1  treated 24hr

SRRxxxx18   0   N1  untreated   12hr
SRRxxxx22   3   N1  treated 12hr
SRRxxxx23   0   N1.1    treated 12hr
SRRxxxx19   3   N1.1    untreated   12hr
SRRxxxx20       N1.1    treated 6hr
SRRxxxx21       N1.1    treated 12hr

can you please suggest my design formula for this. basically there is untreated and treated for each group with multiple replicates. any help is much appreciated.

[rpolicastro]

Do you want to compare untreated to treated regardless of time, or taking into account time?

[Rajesh]

I want to compare respective untreated vs treated, for ex. SRRxxxx13 untreated vs SRRxxxx14 treated. may be i should remove this time point column if that makes these comparisons streamlined for DESeq object. alternatively i was also thinking grouping them in to 1,2,3,4 ... conditions and call DESeq but am not sure if that will give me a comparison result when extracting results using contrast ?

Answer 1

[Mensur Dlakic]

There are several perl scripts here that will help you automate your process. They support DESeq2, edgeR and limma.

[Rajesh]

awesome, thank you! I will check that out

[Rajesh]

in run_DE_analysis.pl I am getting this - Error, cannot determine column index for replicate name [conditionA_rep1]$VAR1 = { .... other conditions here....};

how do i get pass this ?

[Mensur Dlakic]

It may help if you show the first couple of lines from your input files. Chances are that either a file with sample names or the one with counts is not formatted properly. It is all described in that Wiki page, just follow the instructions.

My guess is that you have a name for your first column in counts file, and it shouldn't be there. The first column should have gene/transcript names, followed by tab-delimited columns with counts. Yet on the very first line (first row), the name for gene/transcript column must be empty, while only sample columns are named. It is tough to troubleshoot like this, so I suggest again to read the directions carefully and get in touch with the original authors if needed.

[Rajesh]

it works now, Thanks a ton!

[Mensur Dlakic]

Good to hear. It would be nice if you shared what was wrong as it may help others reading this post later. Also, you may want to consider accepting the answer if it solved your problem.

[Rajesh]

it was the formatting error, as you mentioned. i had wrong col names in my matrix, also had special charecters in names which got corrected and it works. Thanks!