Question

is it possible to create this comparison?

0

Entering edit mode

hamaor • 0

@hamaor-9799

Last seen 5.7 years ago

i'm using deseq2 for my RNASeq analysis.

i have 2 factors, each has 2 levels:

1. genotype : col (wildtype), M6_5 (mutant)

2. treatment: Nacl, control

my design is: ~ genotype + treatment + treatment:genotype.

to get the effect of the treatment for each genotype i used:

results(dds, contrast = c("treatment","Nacl","control")

results(dds,list(c("treatment_Nacl_vs_control","genotypeM6_5.treatmentNacl"))

and to test differences between genotype on the same treatment conditions i used:

results(dds, contrast = c("genotype","M6_5","col")

results(dds, list(c("genotype_M6_5_vs_col","genotypeM6_5.treatmentNacl" ))

now, i'm looking for the comparison : M6_5 Nacl vs col control and M6_5 control vs col Nacl

is that possible using my design?

thanks a lot

rnaseq pairwise multiple comparisons deseq2 • 1.5k views

ADD COMMENT • link updated 8.7 years ago by Michael Love 42k • written 8.7 years ago by hamaor • 0

score 0 · Answer 1 · 2016-03-02

0

Entering edit mode

chimeric • 0

@chimeric-9840

Last seen 7.2 years ago

Hi hamaor, I've found the following solution to be the most intuitive way to test whatever combination of factors and levels you're looking for. The design from this example also looks similar to yours.

A: DESEq2 comparison with mulitple cell types under 2 conditions

Cheers,

Erin

ADD COMMENT • link 8.7 years ago chimeric • 0

0

Entering edit mode

hey Erin,

it seems that it answers the pairwise comparisons of the effect of the treatment for each genotype, or the genotype gene expression on the same treatment.

what i'm trying to do is to mix between different levels of each factor. so if i have genotype I,II and treatment A,B i'm looking for the comparison: II A / I B and II B / I A.

ADD REPLY • link 8.7 years ago hamaor • 0

0

Entering edit mode

Hi hamaor, I think this would still work for you--see the section of the vignette Michael referenced for more info.

To explain, after creating the a "group" factor (dds$treatment and dds$genotype are concatenated to make 1 factor with 4 levels), you can tell the contrasts to run whatever combination of levels you prefer (colcontrol, colNaCl, M6_5control, M6_5NaCl). In this case, your design is ~ group and you'd generate the contrasts you want by running:

results(dds, contrast=c("group","M6_5control","colNaCl"))
results(dds, contrast=c("group","M6_5NaCl","colcontrol"))

ADD REPLY • link 8.7 years ago chimeric • 0

score 0 · Answer 2 · 2016-03-02

0

Entering edit mode

Michael Love 42k

@mikelove

Last seen 13 hours ago

United States

Take a look at the most recent vignette which goes with the current release v1.10. We show a simpler design for extracting pairwise comparisons, instead of using an interaction term. Search for "interactions"

ADD COMMENT • link 8.7 years ago Michael Love 42k