gravatar for Ridha

3 hours ago by

Hello everyone,

I’m doing differential expression analysis using DESeq2 of 6 heart cells(iPSC-CM) that were either treated with iron or not (treatment). The issue is that these 6 samples were taken from 3 different individuals and I am not interested in differences between individuals(sample), but the effect of iron on these heart cells. From the PCA plot, however, it seems that the samples are not separated by the intervention(FE vs AA), but rather by the sample type( despite that I have controlled for the samples in my design(see code below).
From my simple understanding of DESeq2, the code I wrote indicates that the effect of treatment will be measured correcting for the sample type and that's what I want. Even if I switch the design, I still get the same clustering pattern. Am I missing something?
I don't think it's because of the data because another study using the same dataset found clustering based on treatment and not individuals.

Thank you very much in advance for your help!

#Overview of the count data
      X273.AA X273.FE X15.AA X15.FE X2.AA X2.FE
DDX11L1         0       0      0      0     0     0
WASH7P         46      56     64     79    69    68
MIR6859-1       2       0      2      0     2     2
MIR1302-2       0       0      0      0     0     5
FAM138A         0       0      0      0     0     0
OR4G4P          0       0      0      0     0     0

Making my metadata

                     "FE","AA","FE")) # AA= control, FE= intervention
cell_type<- as.factor(c("HiPSC273","HiPSC273","HiPSC15",
          "HiPSC15","HiPSC2","HiPSC2")) # indicating the sample type(from which individual it came from)
# matching the columnames of count  data to rownames of metadata
#### checking #### 
all(rownames(metadata)==colnames(df_io)) # was true

Now making the model and extracting normalized counts to make PCA plot

df_deseq2_ijzer<-DESeqDataSetFromMatrix(countData = df_io,
                                     colData = metadata,
                                     design = ~cell_type+treatment) 

qc_log_ijzer2<-varianceStabilizingTransformation(dd_normalizated,blind = T)

Making the PCA plot



modified 1 hour ago



3 hours ago


Source link