RUVSeq DE analysis for data with replicate samples and 4 conditions

[sharvari14]

Hello,

I have a question on using RUVSeq for removing batch effects for 8 samples of RNA-Seq data - 2 batches with 4 samples each.

The example data in the tutorial shows 2 conditions with 3 replicates each:
https://www.bioconductor.org/packages/release/bioc/vignettes/RUVSeq/inst/doc/RUVSeq.pdf

Following the steps in section 3 to estimate factors of unwanted variation using replicate samples, I am trying to define the matrix as:
differences <- matrix(data=c(1:2,3:4,5:6,7:8), byrow=TRUE, nrow=2)
PCA plot for set3 doesn't group the replicates:

However, if I define matrix for pairwise condition as:

differences1 <- matrix(data=c(1:2,3:4), byrow=TRUE, nrow=2)
set3_1 <- RUVs(set, genes, k=1, differences1)
plotPCA(set3_1, col=colors[x], cex=1.2)
and
differences2 <- matrix(data=c(5:6,7:8), byrow=TRUE, nrow=2)
set3_2 <- RUVs(set, genes, k=1, differences2)
plotPCA(set3_2, col=colors[x], cex=1.2)

The PCA plot for each shows grouping of replicates for the 1st 2 conditions.

Could you please let me know what is the correct approach to remove batch effects from replicates and do DE analysis for pairs of conditions : Control vs. KD and Control75 vs. KD75

set
SeqExpressionSet (storageMode: lockedEnvironment)
assayData: 18201 features, 8 samples
element names: counts, normalizedCounts, offset
protocolData: none
phenoData
sampleNames: Control_1 Control_2 ... KD75_2 (8 total)
varLabels: x
varMetadata: labelDescription
featureData: none
experimentData: use 'experimentData(object)'
Annotation:
pData(set)
x
Control_1 Control
Control_2 Control
KD_1 KD
KD_2 KD
Control75_1 Control75
Control75_2 Control75
KD75_1 KD75
KD75_2 KD75

Thanks for all the help.
Sharvari

[sharvari14]

Correction:

set
SeqExpressionSet (storageMode: lockedEnvironment)
assayData: 18201 features, 8 samples
element names: counts, normalizedCounts, offset
protocolData: none
phenoData
sampleNames: Scrambled_1 Scrambled_2 ... CNOT7KD75_2 (8 total)
varLabels: x
varMetadata: labelDescription
featureData: none
experimentData: use 'experimentData(object)'
Annotation:
pData(set)
x
Scrambled_1 Scrambled
Scrambled_2 Scrambled
CNOT7KD_1 CNOT7KD
CNOT7KD_2 CNOT7KD
Scrambled75_1 Scrambled75
Scrambled75_2 Scrambled75
CNOT7KD75_1 CNOT7KD75
CNOT7KD75_2 CNOT7KD75

[dtenenba]

@sharvari14 it doesn't sound like you are reporting an issue with the package. You should be asking on the support site: https://support.bioconductor.org/ . Be sure and tag your question with the package name (RUVSeq).

[sharvari14]

Sorry about that.

I just posted the question in support.bioconductor.org..

Would appreciate any help.

thanks
Sharvari

[drisso]

Yes, please, post your question on the bioconductor support site.

[sharvari14]

Hello Davide,

I did post my question on Bioconductor. Would really appreciate if someone
could help me with RUVSeq DE analysis.

Thanks
Sharvari

On Fri, Aug 28, 2015 at 2:38 PM, Davide Risso [email protected]
wrote:

Yes, please, post your question on the bioconductor support site.

—
Reply to this email directly or view it on GitHub
drisso/RUVSeq#7 (comment).

[sharvari14]

Sorry, just saw your post.
Thanks

On Tue, Sep 1, 2015 at 10:14 AM, sharvari gujja [email protected] wrote:

Hello Davide,

I did post my question on Bioconductor. Would really appreciate if someone
could help me with RUVSeq DE analysis.

Thanks
Sharvari

On Fri, Aug 28, 2015 at 2:38 PM, Davide Risso [email protected]
wrote:

Yes, please, post your question on the bioconductor support site.

—
Reply to this email directly or view it on GitHub
drisso/RUVSeq#7 (comment).