Sleuth output was converted to matrix format, transposed (to fit Prcomp format) and filtered (sum of gene expression across all samples had to be above 5 tpm)
so <-readRDS("//jic-hpc-data/Research-Groups/Cristobal-Uauy/Anna/sleuth_clipNclean_rds")
matrx <- sleuth_to_matrix(so, "obs_norm", "tpm")
mx <- t(matrx)
length(colnames(mx))
## [1] 298774
mx2 = mx[,colSums(mx) > 5]
length(colnames(mx2))
## [1] 103862
I removed F4, H2 and G2, because they were extreme outliers in PCA graphs
length(rownames(mx2))
## [1] 27
length(rownames(G2))
## [1] 24
G2.pca <- prcomp(G2, center = TRUE, scale. = TRUE)