-2

I am trying to plot a dose response curve using drc. I've got two cell lines (gr, ctrl) and I am trying to figure out the ED50 for both of them in the same plot. The Cells are treated with the same drug, confluence beeing the meassured response.

After calculating using ED the ouput gives me some strange numbers.

> #data
> dput(incu)
structure(list(cell_line = c("gr", "gr", "gr", "gr", "gr", "gr", 
"gr", "gr", "gr", "ctrl", "ctrl", "ctrl", "ctrl", "ctrl", "ctrl", 
"ctrl", "ctrl", "ctrl"), concentration = c(0, 0.001, 0.003, 0.006, 
0.01, 0.03, 0.1, 0.3, 1, 0, 0.001, 0.003, 0.006, 0.01, 0.03, 
0.1, 0.3, 1), confluency = c(100, 83.920208263529, 76.3771201517205, 
70.3196127286416, 87.6047426653901, 73.0547963016434, 86.3958825883478, 
75.3635915616798, 85.5326850667842, 100, 84.6549611260149, 58.9885983810855, 
-8.76517036651747, 21.6983416532331, -10.9003403181362, -12.3609052077884, 
-13.2655463052258, -7.59527744074285), SD = c(NA, 16.4686274085642, 
14.9385036068391, 9.55320006142224, 10.183898021169, 20.001799498955, 
7.22345139045768, 22.0994381721301, 8.43478081150139, NA, 9.63266009058469, 
10.3368371037371, 16.992025977207, 24.2143931244896, 8.20169041108195, 
7.8325260302496, 5.26949485770548, 3.84706305611308)), row.names = c(NA, 
-18L), class = c("tbl_df", "tbl", "data.frame"))
> 
> # fitting
> all.L4 <- drm(confluency ~ concentration, cell_line, data = incu, fct  = L.4())
> 
> # plot
> plot(all.L4, xlab = "Drug Concentration", ylab = "% of Control",  col = TRUE)
> 
> #calculate ED50
> ED(all.L4, 50)

Estimated effective doses

            Estimate Std. Error
e:ctrl:50 -0.0226138  0.0010383
e:gr:50   -0.4515882  0.9800067

enter image description here

My questions:

1: Did I miss anything to calculate the ED50. Is there a way to display the EC50 values as lines in the plot?

Beeing a novice, I much appreciate any advice.

Cheers!

inc
  • 21
  • 5

1 Answers1

1

The question is trying to fit a curve whose Y values are negative with a function that can never give negative values. Try LL.4 instead.

Also the gr cell line looks like a horizontal line so you are not going to get anything reasonable from a logistic or any common function (other than a horizontal line) for that and you can't expect to use ED on such a curve.

G. Grothendieck
  • 254,981
  • 17
  • 203
  • 341
  • Thank your very much. Using `LL.4 ` I got a realistic value of 0.003 uM. I'm not so much concerned about the gr values since those are resistent cells and there should be no response. I'm still wondering if there is a way to "normalize" the data another way in order to avoid negative values? – inc Sep 23 '19 at 14:18