[R] Need to replicate Boltzman Signmodial Curve fit from Graph Pad

David Carlson dcarlson at tamu.edu
Wed Apr 24 18:17:27 CEST 2013


Sorry. I left out a line:

> pHvals <- seq(min(dta$pH), max(dta$pH), length.out=100)

Should be inserted just before the lines() function.

-------------------------------------
David L Carlson
Associate Professor of Anthropology
Texas A&M University
College Station, TX 77840-4352

From: J J [mailto:rnoobier at gmail.com] 
Sent: Tuesday, April 23, 2013 6:49 PM
To: David Carlson
Subject: Re: [R] Need to replicate Boltzman Signmodial Curve fit from Graph
Pad

Thank you David! That looks really close, I believe Graphpad gave a V50 of
 7.244.

One thing I missed on your example was where you obtained 'pHvals' from?


On Tue, Apr 23, 2013 at 5:17 PM, David Carlson <dcarlson at tamu.edu> wrote:
You may want to compare R's four point logistic with the results you get
using Graphpad. For these data the Bottom parameter is not significantly
different from zero (so the three point sigmoid might be adequate), but as
Bert points out, especially with only 8 data points, it is probably
overfitting the model.

> dta <- read.table(text="pH counts
+ 3.8 968
+ 5.0 1347
+ 5.8 2867
+ 6.6 9203
+ 7.0 15817
+ 7.4 20297
+ 8.2 31916
+ 9.2 35756",
+ header=TRUE)
> Sig.nls <- nls(counts~SSfpl(pH, Bottom, Top, V50, Slope), dta)
> summary(Sig.nls)

Formula: counts ~ SSfpl(pH, Bottom, Top, V50, Slope)

Parameters:
          Estimate  Std. Error t value     Pr(>|t|)
Bottom   718.76871   579.25327   1.241     0.282463
Top    36983.98871  1008.31727  36.679 0.0000032986 ***
V50        7.24930     0.05054 143.436 0.0000000142 ***
Slope      0.55582     0.05023  11.065     0.000379 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 767.1 on 4 degrees of freedom

Number of iterations to convergence: 0
Achieved convergence tolerance: 0.000003808
> options(scipen=5)
> confint(Sig.nls)
Waiting for profiling to be done...
                2.5%         97.5%
Bottom  -963.4106068  2243.7154750
Top    34499.9962176 40150.4114887
V50        7.1173103     7.4033391
Slope      0.4382893     0.7114438
> coef(Sig.nls)
       Bottom           Top           V50         Slope
  718.7687104 36983.9887074     7.2493032     0.5558236
> plot(dta, pch=16)
> lines(pHvals, predict(Sig.nls, data.frame(pH=pHvals)), col="red")

-------------------------------------
David L Carlson
Associate Professor of Anthropology
Texas A&M University
College Station, TX 77840-4352

-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] On
Behalf Of Bert Gunter
Sent: Tuesday, April 23, 2013 12:04 PM
To: John Kane
Cc: r-help at r-project.org; J J
Subject: Re: [R] Need to replicate Boltzman Signmodial Curve fit from Graph
Pad

I think you'll find that this is just an alternate parameterization of a 4 P
logistic.
... and in any case there is no statistically detectable difference between
the two.

That is, this is just basically a Graphpad marketing ploy. Fit such data
with anyone's logistic function ... and that is probably overfitting in many
cases, anyway.

Cheers,
Bert

On Tue, Apr 23, 2013 at 9:54 AM, John Kane <jrkrideau at inbox.com> wrote:
> No idea of the area but does this link help?
> http://bioinformatics.oxfordjournals.org/content/24/13/1549.full
>
> John Kane
> Kingston ON Canada
>
>
>> -----Original Message-----
>> From: rnoobier at gmail.com
>> Sent: Tue, 23 Apr 2013 10:55:59 -0400
>> To: r-help at r-project.org
>> Subject: [R] Need to replicate Boltzman Signmodial Curve fit from
>> Graph Pad
>>
>> Hello useRs (please don't kill me),
>>
>> I've fairly new to R having only a few months of playing around with R.
>> What little I've learned has been extremely useful.
>>
>> If someone could point me as to how to replicate the Boltzman
>> Sigmodial curve fit as provided by Graphpad software I'd be eternally
grateful.
>>
>> Where  we currently use Graphpad for only this one function,its seems
>> highly inefficient for a $100 piece of software to be used for only
>> one function (which isn't nearly as bad as the company's pandemic
>> reliance on Excel for nearly everything else).
>>
>> so anyway, what I'd normally do is take a set of data like this:
>>
>>  pH counts  3.8 968  5.0 1347  5.8 2867  6.6 9203  7.0 15817  7.4
>> 20297
>> 8.2
>> 31916  9.2 35756
>> then fit this to a Boltzman sigmodial in Graphpad. Graphpad spits out
>> a much longer set of vectors and also information about v50 and
>> confidence intervals etc (if anyone is familiar with that software).
>> This is what i'd like to replicate.
>>
>> It might be that I just don't know what i'm doing,so feel free to
>> call me an idiot but any help is greatly appreciated!
>>
>> -JJ
>>
>>       [[alternative HTML version deleted]]
>>
>> ______________________________________________
>> R-help at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide
>> http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>
> ____________________________________________________________
> FREE 3D EARTH SCREENSAVER - Watch the Earth right on your desktop!
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.



--

Bert Gunter
Genentech Nonclinical Biostatistics

Internal Contact Info:
Phone: 467-7374
Website:
http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biost
atistics/pdb-ncb-home.htm

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.



More information about the R-help mailing list