Problem while running ballgown package

[Pallavimore]

I am using a ballgown R package to Run the differential expression analysis. Following is my pheno data example. I have two conditions and 3 replicates.

ids      Treatment cell
sample1  drug1   wt1C7
sample2  drug1   wt1C7
sample3  drug1   wt1C7
sample4  drug2   pv1ko
sample5  drug2   pv1ko
sample6  drug2   pv1ko

pheno_data = read.csv("phenodata.csv")
bg = ballgown(dataDir = "ballgown", samplePattern = "sample", pData=pheno_data)
bg_filt = subset(bg,"rowVars(texpr(bg)) > 1",genomesubset=TRUE)
results_transcripts = stattest(bg_filt, feature="transcript", covariate="cell", adjustvars = c("Treatment"), getFC=TRUE, meas="FPKM")

I am getting an error when I run the following command.
results_transcripts = stattest(bg_filt, feature="transcript", covariate="cell", adjustvars = c("Treatment"), getFC=TRUE, meas="FPKM")

Error message :
Error in solve.default(t(mod) %*% mod) :
system is computationally singular: reciprocal condition number = 7.38788e-28
In addition: Warning message:
Partial NA coefficients for 20083 probe(s)

Could anyone please help me to solve the above error?

[alyssafrazee]

Hi there, This error is happening because you're using "cell" as your covariate and adding "treatment" as an adjustment variable, but the structure of your data means you can't look at both cell and treatment here because treatment is collinear with cell: you've treated all your wt1C7 cells with drug1, and all your pv1ko cells with drug2. Adjusting by treatment means you want to know the difference in expression between cell types on the same drug, but if you filter your data set to drug1 you only have wt1C7 cells, and if you filter your data set to drug2 you only have pv1ko cells. You can fix the error by not specifying an adjustment variable, though you won't be able to tell if any effects you find are due to the treatment or the cell type (since they are totally intermingled). This is because of the experimental design, and it will be true for ballgown or any other software you might use. Hope this helps! -Alyssa

…

On Wed, Oct 23, 2019 at 1:07 PM Pallavimore ***@***.***> wrote: I am using a ballgown R package to Run the differential expression analysis. Following is my pheno data example. I have two conditions and 3 replicates. ids Treatment cell sample1 drug1 wt1C7 sample2 drug1 wt1C7 sample3 drug1 wt1C7 sample4 drug2 pv1ko sample5 drug2 pv1ko sample6 drug2 pv1ko pheno_data = read.csv("phenodata.csv") bg = ballgown(dataDir = "ballgown", samplePattern = "sample", pData=pheno_data) bg_filt = subset(bg,"rowVars(texpr(bg)) > 1",genomesubset=TRUE) results_transcripts = stattest(bg_filt, feature="transcript", covariate="cell", adjustvars = c("Treatment"), getFC=TRUE, meas="FPKM") I am getting an error when I run the following command. results_transcripts = stattest(bg_filt, feature="transcript", covariate="cell", adjustvars = c("Treatment"), getFC=TRUE, meas="FPKM") Error message : Error in solve.default(t(mod) %*% mod) : system is computationally singular: reciprocal condition number = 7.38788e-28 In addition: Warning message: Partial NA coefficients for 20083 probe(s) Could anyone please help me to solve the above error? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#164?email_source=notifications&email_token=AAPJ5WCPRXJX7DBAPGAU3CLQQA46ZA5CNFSM4JEAJ4LKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HTZPW4A>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAPJ5WBXRQAO7FV3S374OATQQA46ZANCNFSM4JEAJ4LA> .

[Pallavimore]

Thank you so much for the explanation. It worked now.