I am currently using R to assign mutation significance values to proteins in yeast.
I have a data frame that looks like this:
Genes q_values
1 HNT1 4.836462e-01
2 EMP47 6.792469e-01
3 QDR2 6.357284e-01
4 TMS1 9.781394e-01
5 TMS1 8.672664e-01
...
However, on occasion there will be a q_value significantly lower than the others:
...
35 HHF1 5.565396e-01
36 RGA2 2.323061e-12
37 CDC24 8.174687e-01
...
# Notice how value for row 36 is very low
I want to rescale these q_values onto a 1-10000 scale. However, I need the highest original q_value (i.e. ~9.85e-01) to be the lowest on the new scale (e.i. a value of 1). Inversely, the lowest original q_value (i.e. ~1.36e-13) needs to be the highest on the new scale (e.i. a value of 10000).
I have tried making a variation of the equation proposed here: https://stats.stackexchange.com/questions/25894/changing-the-scale-of-a-variable-to-0-100. However I did not manage to get the results I am looking far.
What would be the best way to go about doing this?
Maybe you can try the code below for rescaling the q values
within(df, rescaled_q_values <- 1e5*(max(new_q_values)-new_q_values)/diff(range(new_q_values)))
which gives
new_yeast_genes new_q_values rescaled_q_values
1 HNT1 4.836462e-01 50554.47
2 EMP47 6.792469e-01 30557.25
3 QDR2 6.357284e-01 35006.36
4 TMS1 9.781394e-01 0.00
5 TMS1 8.672664e-01 11335.09
35 HHF1 5.565396e-01 43102.22
36 RGA2 2.323061e-12 100000.00
37 CDC24 8.174687e-01 16426.16
Data
df <- structure(list(new_yeast_genes = c("HNT1", "EMP47", "QDR2", "TMS1",
"TMS1", "HHF1", "RGA2", "CDC24"), new_q_values = c(0.4836462,
0.6792469, 0.6357284, 0.9781394, 0.8672664, 0.5565396, 2.323061e-12,
0.8174687)), class = "data.frame", row.names = c("1", "2", "3",
"4", "5", "35", "36", "37"))