[RASMB] radial scans 6 sector cells

[Borries Demeler]

Seems like I forgot the attachment, here it is again incl. the attachment...

******************************************************************************

Interesting discussion...

In my experience the conservation of signal approach for constraining
equilibrium fits is only marginally useful at best. I have never been
happy with the results and too many uncertainty factors contribute to
this constraint, and therefore it really doesn't provide the desired
accuracy. These are among the factors you need to consider:

1. you need a VERY accurate position of the bottom of the cell, because
just a tiny inaccuracy of your bottom position makes a large difference
in the signal because of the exponential shape of the curve, which is
largest at the bottom, especially for higher speeds & sigmas. Since you
cannot fit the bottom (it would be correlated with your constraint),
you need to measure it. It is often impossible to tell just from the 
absorbance or interference trace where exactly the bottom position is.
Moreover, due to rotor stretching each speed will produce a different
meniscus and bottom position, so you cannot average between speeds.

2. Nonlinearity in the absorbance optics at larger ODs will give you
errors in the region where you need the most accuracy.

3. the radial precision of the Beckman absorbance scanner is lousy.
This problem does not exist when you use interference optics, but
there the exact meniscus position is difficult to pin down. In 
fluorescence experiments the bottom of the cell is completely 
obliterated from the angle of the laser, and hence not useful.

4. At the bottom of the cell, especially at higher speeds, the gradient
can produce high enough concentrations to where you favor the production
of aggregates, which simply pellet/drop out of solution, and no longer
contribute to the overall signal.

You would have to control ALL of these conditions correctly before you
can have any hope that conservation of signal would be useful.

I don't know of any cure for (4), while (2) and (3) can be alleviated
by using interference optics. To deal with (1), we do the following
in UltraScan: I carefully scan the geometry and dimensions of each
type of centerpiece we use on a high-resolution flatbed scanner and
use gimp (www.gimp.org) to carefully measure the dimensions for each
channel. These are entered into the UltraScan database and associated
with each measurement. Then we calibrate the rotor stretch of each rotor,
where we actually calibrate the rotor stretch with both titanium and epon
centerpieces (because of the difference in the weight), and fit it to a
2nd order polynomial of the rotor speed. Together with the centerpiece
measurements, UltraScan calculates a theoretical bottom of each cell
depending on speed, which in my experience is about as close as you can
reasonably get. For the rotor calibration, I fill the rotor with empty
cells, plus the counterbalance and scan the entire radial domain in
intensity mode to see where the channel starts and ends. I average the
radial points in the steep region at the bottom and the top and measure
how much the radius shifts with each 1000 rpm increment. The shifts
from all channels, top and bottom, are averaged for best statistical
determination to get an accurate rotor stretch. This calibration should
be done for each rotor individually and should be repeated as the rotor
ages. The attached image shows a screenshot of the UltraScan-3 utility
that processes the calibration data.

I think if you do all of this you can get a fairly reliable value for
the bottom of your cell position, although it doesn't take into account
slack in the cell housings that may distort the predicted value somewhat.

So in theory, I agree with Walt, but in practice I find John's method
is about as useful as any other approach. In my opinion, given the
excellent methods available now for whole boundary fitting of velocity
data, which can also give you Kds and molecular weights, and even slow
kinetics, I rely mostly on velocity anymore, because I find the results
more reproducible and convincing.

That's about it for my $0.02...

cheers, and happy holidays to all!
-Borries
-------------- next part --------------
A non-text attachment was scrubbed...
Name: rotor_calibration.png
Type: application/octet-stream
Size: 79697 bytes
Desc: PNG image data, 966 x 640, 8-bit/color RGB, non-interlaced
URL: <http://list.rasmb.org/pipermail/rasmb-rasmb.org/attachments/20101213/4de67499/attachment.obj>