[R] reproducing Graphpad IC50 in R with drc package

[Jon Arsenault]

I've been banging my head against the wall a bit with this and would be
ecstatic if someone could help.

Initial data:

     Response Dose
1  285.17      0.125
2  377.65      0.250
3  438.99      0.500
4  338.46      1.000
5  227.87      2.000
6  165.96      0.010
7  302.92      0.125
8  418.50      0.250
9  464.69      0.500
10 301.36      1.000
11 213.12      2.000
12 160.34      0.010
13 306.18      0.125
14 435.37      0.250
15 451.34      0.500
16 319.50      1.000
17 219.83      2.000
18 172.52      0.010
19 306.56      0.125
20 439.01      0.250
21 469.74      0.500
22 318.05      1.000
23 223.09      2.000

The graphpad template that that this normally would be placed in then
transforms the taking the log(Dose,base=10) and normalizes the Response.
I've confirmed that I was able to recreate that here:

   NormResponse  LogDose
1        41.011 -0.90309
2        72.909 -0.60206
3        94.067 -0.30103
4        59.392  0.00000
5        21.246  0.30103
6        -0.108 -2.00000
7        47.133 -0.90309
8        87.000 -0.60206
9       102.932 -0.30103
10       46.595  0.00000
11       16.159  0.30103
12       -2.047 -2.00000
13       48.258 -0.90309
14       92.819 -0.60206
15       98.327 -0.30103
16       52.852  0.00000
17       18.473  0.30103
18        2.155 -2.00000
19       48.389 -0.90309
20       94.074 -0.60206
21      104.674 -0.30103
22       52.352  0.00000
23       19.598  0.30103




Now graphpad used 'log(inhibitor) vs. normalized response -- Variable
slope' which it states to be a 4parameter sigmodial but it gives me very
different results.


graphpad gives and IC50 of 0.1560 and drm's LL.4 gives 0.1149 and it wont
even take the transformed data,just throws an error.

	[[alternative HTML version deleted]]