First of all, I would thank the organizer for giving this opportunity for me to present

here.

I'm talking about the programs in Ultrascan.

We usually use to analyze multi-wavelength data in analysis of the

scan model to calculate the pseudo-absorbance.

And finally, I'm talking about the multi-wavelength profile decomposition and the error produced

during the process of the decomposition.

And finally, profile normalization to compare different component profiles we get from the

decomposition procedure.

Each experiment has multiple noises.

Each noise, the top one has all types of the noises.

During the data modeling, we get very nice and clean data which matches the LAM equations.

We remove all of these types of noises during the data modeling.

One of the most important noises we see in the experimental data is the time-imbrane

noise.

Each radial point has the offset that is constant over time.

And during the modeling, it is removed.

But there is another type of the noise as we call it, radially-imbrane noise.

It is the fluctuation of the xenon lamp over time.

The final noises involved in each experiment is the stochastic noise.

So we cannot remove it from the final model.

These two types of noises, the time-imbrane noise and the radially-imbrane noise, are

clear from the data modeling.

So we can see the procedure.

We can do that.

During the data modeling, especially in the velocity sedimentation experiments, what you

are going to do is to minimize these equations.

We have experimental data and a model coming from the LAM equations.

And you can see that before we start to do that, we can remove the radially-imbrane noise

by simply integrating the area of each scan and feed a polynomial data and remove that

type of error.

And then the only things that remain are the time-imbrane noise and the stochastic noise.

So in an iterative approach, we try to get a better model and by increasing the accuracy

of the model calculations, we can finally estimate a better time-imbrane noise.

But you can see that there is always a side effect of the stochastic noise in the data

modeling that by increasing the number of scans, you can decrease this side effect and

finally find the time-imbrane noise.

It is the regular approach we use in the sedimentation velocity experiment.

But in the ABD, because we do not have any model to use the same procedure to find the

time-imbrane noise, you have to follow different methods to calculate the absorbance data.

One solution is doing the experiment in absorbance mode.

But you know that by doing the absorbance mode, we increase the stochastic noise by

the factor of sqrt of 2.

There is another issue that I'm going to talk about more in the next slides.

There is always some time-imbrane noise coming from the window of the cell.

And the absorbance mode cannot take over that.

What we did was to split the time-imbrane noise into two big terms of the time-imbrane

noise coming from the optics, the photomultiplier system and the mirrors.

And the other one is coming from the window of the cell.

So for calculating the time-imbrane noise from the optics, we scan the whole cell.