search

BIMSBbioinfo / pigx_rnaseq

Bulk RNA-seq Data Processing, Quality Control, and Downstream Analysis Pipeline
GNU General Public License v3.0
20 stars 11 forks source link

Paired analyses? #114

Closed smoe closed 2 years ago

smoe commented 2 years ago

We have several cell lines that we expose to two drugs in all combination, i.e. cell1 w/o all drugs, cell1+drug1, cell1+drug2, cell1+drug1+drug2. And the same for the other cell lines. We have the cell line as a covariate now, but I do not see that any paired analysis would be performed, i.e. the comparison of cell1 data only other data from the same cell line.

Am I correct?

borauyar commented 2 years ago

You can define as many differential expression analyses as you want for any groups of pairs of samples. However, the assumption is that the experimental setting has to follow a "case" vs "control" structure. Any groups of samples can be used as controls and any groups of samples can be used as "case" samples.

While defining this, you need to use the identifiers for samples under "sample_type" column.

Let's say you want to compare the impact of drug1 on cell line 1, cell1: cell line 1 without drug treatment (assuming at least two biological replicates) cell1_drug1: cell line 1 with drug1 (assuming at least two biological replicates)

analysis1:
      case_sample_groups: "cell1_drug1"
      control_sample_groups: "cell1"

You could also group multiple conditions like this: 

analysis2:
      case_sample_groups: "cell1_drug1, cell1_drug2"
      control_sample_groups: "cell1"

So, it is completely up to the user which pairs of groups of samples are to be compared.

smoe commented 2 years ago

Thank you for your quick reply. I cross-checked with old routine in Limma (https://bioconductor.org/packages/release/bioc/vignettes/limma/inst/doc/usersguide.pdf page 43) and I may be closer to what I want than I had thought :) This needs some more time for me to dig into.