s-news
[Top] [All Lists]

NLME

To: <s-news@lists.biostat.wustl.edu>
Subject: NLME
From: Jason Connor <jconnor@bio.ri.ccf.org>
Date: Fri, 22 Nov 2002 11:23:59 -0500 (EST)
Using NLME in version 6.1 on a Sun Sparc.

I get getting a message about NA's even though I have no missing data in
my dataset.

I have 4 pairs of corneas, 1 case - 1 control per pair.
Corneas were stored in solution and concentration of a chemical
was measured everyday for 21 days in all 8 samples.

> noGD
Grouped Data: Conc ~ Day | Sam
   Day Group Eye       Conc Sam
 1   1     0   A 0.00000000  A0
 2   2     0   A 0.10595760  A0
 3   3     0   A 0.53191440  A0
 4   4     0   A 0.90191355  A0
 5   5     0   A 1.13137230  A0
 6   6     0   A 1.65407400  A0
 7   7     0   A 2.03353260  A0
 8   8     0   A 2.21974815  A0
 9   9     0   A 2.45190960  A0

To prove I have no missing:

> apply(is.na(noGD),2,sum)
 Day Group Eye Conc Sam
   0     0   0    0   0

And I have numerics and factors where appropriate:

> unlist(lapply(noGD,class))
       Day     Group      Eye      Conc      Sam
 "numeric" "numeric" "factor" "numeric" "factor"


But when I try to fit the Michaelis-Menten model for each of the 8 corneas
I receive a strange error:

> fitList <- nlsList(Conc ~ SSmicmen(Day, Vm, K), data=noGD)


Problem in nls(y ~ x/(K + x), data = xy, start = lis..: 21 NAs found in
value, while calling subroutine setup_nonlin
Use traceback() to see the call stack

(this error repeats 8 times).

I get the same error when trying to fit 1 model to all 168 data point:

> fit <- nls(Conc ~ SSmicmen(Day, Vm, K), data=noGD)
Problem in nls(y ~ x/(K + x), data = xy, start = lis..: 21 NAs found in
value, while calling subroutine setup_nonlin
Use traceback() to see the call stack

Any suggestions for things to try?

Thank you.
Jason



<Prev in Thread] Current Thread [Next in Thread>
  • NLME, Jason Connor <=