Question: Highly biased experiment showing funny histogram of p values for all tests. Suggestions?
gravatar for Mozart
6 months ago by
Mozart140 wrote:

Dear all, I am struggling (as usual) with a really important/expensive experiment we manage to complete. Having done this in different days, plus given the fact we are not expecting a strong phenotype (we are treating cells with a molecule that shouldn't change these ones that much), I would imagine not so many genes differentially expressed. Perfect. I initially corrected the experiment for batch effect. Indeed, I added an extra column to my model matrix in this way:

Table <- data.frame(condition = factor(c("A","A","A","B","B","B","C","C","C","A","B","C","D","D","D","D")))

Table$batch = factor(c('day1','day2','day3','day1','day2','day3','day1','day2','day3','day4','day4','day4','day4','day1','day2','day3'))


Thus, having:

> table(Table)
condition day1 day2 day3 day4
    A      1    1    1    1
    B      1    1    1    1
    C      1    1    1    1
    D      1    1    1    1


dds <- DESeqDataSetFromTximport(txi.kallisto.tsv, Table, ~batch+condition)
dds <- DESeq(dds)
nrow(dds) **#[1] 29654**
res = results(dds)

Say I want to compare A(treated) vs B(untreated):

dds$condition <- relevel(dds$condition, ref = "B")
dds <- DESeq(dds)

results_A_vs_B<-results(dds, c("condition", "A", "B"))

If I plot the histogram of p values I get something like this: enter image description here Which is pretty odd I guess, given the fact I get a lot of unfiltered values in my final experiment that are around pval=1.

then, I decided to remove low counts by arbitrarily choosing the following filtering parameter:

idx <- rowSums( counts(dds, normalized=TRUE) >= 8 ) >= 6
dds <- dds[idx,]
res = results(dds)


dds$condition <- relevel(dds$condition, ref = "B")
dds <- DESeq(dds)

results_A_vs_B<-results(dds, c("condition", "A", "B"))

it's even worse, I guess:

enter image description here

So there's no doubt there's something wrong here but not sure how to get rid of it.

Do you have any ideas?


filtering rna-seq deseq2 • 224 views
ADD COMMENTlink written 6 months ago by Mozart140

I cannot see the figures, is it me or something went wrong?

ADD REPLYlink written 6 months ago by Fabio Marroni2.5k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 713 users visited in the last hour