ShufflingLabelsForDE

shuffle labels of cases / controls to assess whether pValues are due to chance

files needed

You need an expression matrix and the accompanying phenotype table. Adapt:
1.the labels inside the script to match your phenotype (e.g. Status, Age)
2.the formula for sample design to reflect the main category and covariates

On Rosalind I ran it like this

sbatch -p brc --mem=10G runR.sh A
sbatch -p brc --mem=10G runR.sh B
sbatch -p brc --mem=10G runR.sh C

'A' is a prefix for output file. I did it becasue I split it in 100 random permutations, and the results of each one will be written to a different file, e.g. significantPermutationsA11.41.13.csv. I sent 10 jobs with different prefix letters so that they could run at once.

Merge the output files and look at results

this is also in script countingPermutationsFromCsvRes.R

Go to the directory where all csv files from previous step are

From R:

files <- list.files(pattern = "\\.csv$")
files

In my case I got something like this, a list of all of them:

[1] "significantPermutationsB19.40.57.csv"
[2] "significantPermutationsC19.28.36.csv"
[3] "significantPermutationsD20.02.40.csv"
[4] "significantPermutationsE20.02.39.csv"
[5] "significantPermutationsF20.06.08.csv"
[6] "significantPermutationsG20.05.04.csv"
[7] "significantPermutationsH20.04.42.csv"
[8] "significantPermutationsK19.29.54.csv"
[9] "significantPermutationsL19.55.33.csv"
[10] "significantPermutationsM19.22.38.csv"

sigs<-c(2443,207,1285)

these are hardcoded numbers, the numbers of significant genes at pValue cutoff 0.1, 0.01, 0.0), before the permutations, in the original scenario with real cases vs controls

for (file in files) {
                    sigs_temp <- read.table(file,row.names=1,header=F,sep=",",skip=1)
                    sigs<-cbind(sigs,sigs_temp)
                }
dim(sigs)
#              [1]    3 1001

The following code is just an example of how to explore the numbers in sigs. Each number represents the number of significant genes at either 0.1, 0.01 or 0.05

length(as.numeric(sigs["significant0_05",2:1001])[as.numeric(sigs["significant0_05",2:1001])>1285])
 #             [1] 1
length(as.numeric(sigs["significant0_01",2:1001])[as.numeric(sigs["significant0_01",2:1001])>207])
#              [1] 0
length(as.numeric(sigs["significant0_1",2:1001])[as.numeric(sigs["significant0_1",2:1001])>2443])
 #             [1] 1
summary(as.numeric(sigs["significant0_1",2:1001]))
#              Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
#              0.00    1.00    3.00   19.05    8.00 3562.00 
summary(as.numeric(sigs["significant0_01",2:1001]))
 #             Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
#              0.000   0.000   1.000   2.478   2.000 103.000 
summary(as.numeric(sigs["significant0_05",2:1001]))
 #             Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
#              0.000    1.000    2.000    8.152    5.000 1821.000