How transform FPKM values to Z-score using R

Question

[mjoyraj]

I have FPKM values of 2766 genes for 15 different conditions. I want to transform the FPKM into Z-score. How can I do this?

A small glimpse of the data is given below

Gene                 Cluster   cEB1      cEB2      cEb3      cLB1      cLb2      cLB3      cEF1       cEF2      cEF3       cLF1       cLF2       cLF3       cMF1       cMF2       cMF3
Chr1_FK1             1         32125.1   12358.3   5830      9615.8    21497.2   16153     7301.54    5689.22   1559.81    1046.97    9.70811    12.6835    8767.73    8039.51    11612.4
ENSGALG00000000277   1         9.45172   1.13001   3.33611   8.02945   4.31835   1.73261   0.454218   0         0.595844   0          0          0.139979   0.083132   0          0.08953
ENSGALG00000000341   1         25696.9   5322.23   3800.56   3296.56   9739.64   13518     3600.36    2325.34   62.2555    7.96366    1.94201    8.05702    5092.02    3704.02    6048.07
ENSGALG00000000521   1         141.774   65.2475   54.9138   57.7126   96.6525   97.7173   50.6554    47.9005   35.9634    17.1921    20.2356    27.2313    66.7392    50.5141    61.0671
ENSGALG00000000526   1         3.9706    1.15042   1.63248   1.0424    1.69618   1.69195   1.78895    1.14488   0.591284   0.252008   0.594155   0.442257   1.79887    1.39072    1.30568
ENSGALG00000000645   1         3.20263   2.25025   2.47694   2.8685    2.13484   1.60007   0.801766   1.16636   0.830455   0.054204   0.099204   0.151425   0.424215   0.431372   0.314711
ENSGALG00000000748   1         30.7      17.9308   13.7772   9.71653   22.2964   22.2266   7.1346     4.97026   2.45806    1.12066    1.24514    1.30448    8.35496    5.04605    7.71874
ENSGALG00000001885   1         25.7813   13.5145   21.4617   22.9718   17.574    18.2177   3.42242    5.17301   2.50592    0          0.226905   0.346521   3.44081    2.64345    3.74202
ENSGALG00000002412   1         43.9355   31.886    19.9528   24.7974   28.5074   31.8579   24.0985    21.3538   20.1413    7.21469    6.4085     5.69435    21.5841    18.0618    22.9048
ENSGALG00000002681   1         141.55    81.2838   50.0476   93.3922   76.0349   67.5903   57.3259    87.5868   42.1664    21.2901    15.156     15.5312    53.339     54.5204    66.068
ENSGALG00000002988   1         206.862   110.517   137.667   111.514   72.4366   71.3408   31.216     38.0035   36.928     59.9633    10.1934    19.3801    20.2704    18.8814    65.0859
ENSGALG00000003921   1         312.418   218.574   245.381   176.837   194.555   214.552   83.4999    84.6692   65.0635    124.388    133.099    109.405    155.528    145.834    154.364

Answer 1

[dariober]

The z-score in R code is:

zscore<- function(x){
    z<- (x - mean(x)) / sd(x)
    return(z)
}
z.fpkm<- zscore(data$fpkm)

I would probably log transform the raw fpkm values to make them closer to a normal distribution and consequently make the z-score more meaningful.

[mjoyraj]

What is X and Z here...??

[mforde84]

x is all the data you're using for the population stats. z is holding the calculation, then returning the data to a specified output.

also dariober is correct, you should log transform first. also, you may want to see if you can switch to TPM instead of FPKM or RPKM.

Answer 2

[Sean Davis]

See the R scale() and sweep() R functions.