Question

I have a very huge dataframe (about 700 columns and 4,00,000 rows). Example as follows:

           "samp1"             "samp2"              "samp3"
"ct1"   3.45909539741409    4.16162745114877    4.28378657903742
"ct2"   2.45516883029741    2.64529479763739    2.62023669114738
"ct3"   -1.53506396613939   -1.53318358856732   -1.2235498333358
"ct4"   2.54561049516449    1.91067140586626    2.6860636959124
"ct5"   1.08270893850391    0.847112473835081   0.90241673852809
"ct6"   -1.49057768051391   -1.09321836721649   -1.11635564230449
"ct7"   -0.82888593769947   -0.533641693674542  -0.339345700770896

All the columns contain log2 values in every numeric cell (lets say “a”) and i wish to convert them all to normal values (lets say “b”), with the following formula:

b = 2^a / [(2^a) + 1]

So my resulting file should be like the above, but with all the converted values of b, instead of a.

Can anyone help me with this ?
Thanks in advance.

Answer 1

Create some data

a <- matrix(runif(21), ncol=3,
            dimnames=list(paste0("ct", 1:7), paste0("samp", 1:3)))

a
         samp1     samp2      samp3
ct1 0.36856468 0.1969006 0.21410103
ct2 0.80894310 0.4402108 0.96686681
ct3 0.52169299 0.1410873 0.35899306
ct4 0.75243588 0.5432110 0.85099247
ct5 0.77734048 0.5657070 0.33521023
ct6 0.73011232 0.1539275 0.08615577
ct7 0.08505216 0.4905622 0.13415698

Transform:

b <- 2^a / ((2^a) + 1)

b
        samp1     samp2     samp3
ct1 0.5635223 0.5340674 0.5370329
ct2 0.6366183 0.5756963 0.6615437
ct3 0.5894301 0.5244291 0.5618897
ct4 0.6275099 0.5930347 0.6433337
ct5 0.6315359 0.5967925 0.5578276
ct6 0.6238860 0.5266483 0.5149252
ct7 0.5147341 0.5841982 0.5232309