[R] Help with error

Gretchen Heavner glh33 at cornell.edu
Tue Jun 15 17:55:17 CEST 2010


Hi.  I am trying to do a nonlinear regression on a set of data with Monod
kinetics and Haldane inhibition.  I am using the following commands to do
the nonlinear regression:

dce<-read.delim("data.txt", header = TRUE, sep = "\t", quote="\"", dec=".",
fill = TRUE, comment.char="")
dce.m1<-nls(rate~kmax*conc/(Ks+conc+((conc^2)/Ki),data=dce
,start=list(Ks=1000,kmax=6000,Ki=10000),trace=TRUE)

I get reasonable answers for Ks, kmax and Ki from this.
However, I want to find the 95% confidence intervals for each of these
values.  So I first use the command profile:

(pr <- profile(dce.m1))

And then I ususally use the following command to get the confidence
intervals:

confint(pr, level=0.95)

However, when I run the profile command, I am getting the following error:

Error in prof$getProfile() : singular gradient

Do you have any idea what this means, or what I can do to fix it?
I have attached the corresponding file with the data I'm using in case that
helps.

Thanks!
Gretchen Heavner
-------------- next part --------------
conc	rate
1.642954108	4.460677577
1.96419336	5.369018741
2.411233484	303.4816161
2.083860628	406.5940798
3.237105618	332.5106627
1.541574251	310.1025388
0	2.746097252
0	3.236122145
0	5.387406642
2.613273505	544.3881261
1.674376521	535.2056647
1.272419565	688.773693
6506.124712	3193.54196
636.4890845	534.5243761
179.4848649	642.7148796
394.8780499	536.3999568
821.4336211	3018.968118
1985.399238	2880.47305
203.254026	1030.660214
11.6416869	446.6209406
1159.841416	2240.487362
179.1906363	1245.802977
25.32770902	562.8883166
278.1572449	709.1081229
24.3055219	249.1544286
187.6656906	678.752123
8.881946288	350.5795416
7554.586825	2724.21323
7430.314722	2745.073389


More information about the R-help mailing list