Search code examples
rnormalizationstandardized

R function for normalization based on one column?

Is it possible to normalize this table in R based on the last column(samples) samples = number of sequenced genomes. So I want to get a normalised distribution of all the genes in all the conditions.

Simplified example of my data:

enter image description here

I tried:

dat1 <- read.table(text = " gene1   gene2   gene3   samples 
condition1  1   1   8   120
condition2  18  4   1   118
condition3  0   0   1   75
condition4  32  1   1   130", header = TRUE)

dat1<-normalize(dat1, method = "standardize", range = c(0, 1), margin = 1L, on.constant = "quiet")

But the results include negative values and I am not sure how useful this approach is. Can anyone please suggest how I should normalize my data ... to get meaningful results.

Thanks a lot and apologies if it is a dumb question.

Solution

  • Using your data, you write a min max function first:

    minmax = function(x){ (x-min(x))/(max(x)-min(x))}

    Then iterate through the columns:

    norm = data.frame(lapply(dat1[,1:3],function(i) minmax(i/dat1$samples)))

    And it looks like this, I hope it's correct:

           gene1     gene2      gene3
1 0.03385417 0.2458333 1.00000000
2 0.61970339 1.0000000 0.01326455
3 0.00000000 0.0000000 0.09565217
4 1.00000000 0.2269231 0.00000000