Protein standard curves: linear regression vs. polynomial fit

[Jan van der Lee]

Wolfgang Schechinger wrote:

> Hi all,
>
> I noticed that when doing Bradford-sytle protein concentration
> measurements, a 2nd order fit (y=ax^2+bx+c) repeatedly seems to be
> smoother than a linear fit (y=ax+b) for the standard curve.
>
> Has anybody an explanation for this?
>
> Wolfgang
> -----
> usual disclaimers apply * This message is RNAse free - please don't touch!
> -----
> Wolfgang Schechinger
> University of Tuebingen, Germany
> email: wgschech at med.uni-tuebingen.de * wwWait: http://www.medizin.uni-tuebingen.de/~wgschech/start.htm
> -------

Wolfgang,

The actual fit you should use is y= ax + bxy

or: y= (ax)/(1-bx)

The reason: this reaction is limited in final concentration (starts at 0 and has a maximum eg 2.0 in
infinity)

OD ...

l                            ----------
l                      i
l                  i
l              /
l          /      (example curve)
l       /
l    /
l  /
l /
l/_______________________________________
                                                         (protein conc.)
The curve is only lineair in the lower part.

If you want an/or can use it I could send you a Lotus 1-2-3 spreadsheet with fitting possiblities for all
kinds of curves
Mail me back if you want them.

Regards   Jan (holland)