2.6.3 Gene expression analysis
We identified differentially expressed transcripts (DETs) between
control and Cu-treated plants from Sc3 and Sc4 using edgeR v3.24.3100. First, we filtered low count transcripts by
removing those with less than 2 cpm (counts per million; i.e. less than
10 counts per transcript) in at least 3 samples to ensure that any
transcript that was highly expressed in all three replicates from one
group was kept. A total of 23,016 transcripts (7% of the total) passed
this filter and were used in subsequent steps. Second, we fitted the
model using the function glmQLFit (option robust = TRUE) and
tested for DETs using the function glmTreat . Transcripts were
considered differentially expressed when the FDR < 0.001 and
the expression change was ≥ 4-fold (log2FC ≥ 2).
To better understand the changes associated with Cu exposure, we
performed a Fisher’s exact test (with the FatiGO package101 implemented within OmicsBox) on the list of DETs
from each population. With this method, we tested for significant
differences in the fraction of transcripts annotated with a specific GO
term between the DETs list (test set) and the full list of annotated
transcripts (reference set). We applied an FDR cutoff of 0.01, and used
the “Reduce to most specific ” option within OmicsBox to remove
the more general, less informative, GO terms.