FW: [RASMB] Origin software and vbar

[Borries Demeler]

> 
> James Cole wrote:
> This is a really important discussion for the group. As someone 
> developing a new equilibrium package (it is called HeteroAnalysis, more 
> info to follow soon), I'd be very interesting in hearing from more of 
> the RASMB members regarding what packages they are using for analysis of 
> equilibrium experiments, what feature they like/dislike, and most 
> importantly, what features they wish were available.  Regarding the 

Jim,

For what it's worth, here are my $0.02...

here is a list of requirements I have, in no particular order:

1. convenient editing of data limits during analysis
2. a choice of fitting algorithms
3. predefined and user-defined models, incl. heteroassociation
4. platform portability (Linux, Unix, Windows, Mac OSX)
5. a convenient user interface
6. automatic corrections for vbar (from peptide files), density, extinction (from 
buffer component database)
7. fitting of wavelength scans to correct multi-wavelength data extinction coefficients
8. unlimited number of scans to be included in a global fit
9. MySQL database backend
10. reporting the fitting equation used for a particular model
11. reporting the molecular weight
12. allowing the user to set whether the association constant, the 
dissociation constant or log of association constant is reported
13. user-modifiable graphics and plot customizations a al Origin for publication
quality output 
14. Monte Carlo capability with cluster interface to check reliability of fit
15. a good user manual with online, context specific help
16. automatic initialization of fitting parameters
17. report ln(C) vs r^2, MW vs. r^2, MW vs. C, C(MW), residuals, overlays, and concentration
histograms in OD and molar units
18. show oligomer distributions in reversibly self-associating systems
19. For the novice: some check to see if the included data is useful/appropriate
for fitting!
20. multi-threaded/parallelized optimization routines to take advantage of
multi-processor machines
21. presentation module for results (HTML reports, web interface)
22. an experimental design module that suggest the best speeds for the 
experiment and gives you some sort of idea how long the experiment might take 
by simulating it (multiple speeds of course).
23. knowledge about the centerpiece geometry (from centerpiece database)
24. knowledge about the rotor stretching (from rotor calibration database)
25. ability to exchange results/data with collaborators through the internet
26. fitting of global data regardless from which optical system
27. combine multiple experiments into a global fit
28. ability to apply individual buffer corrections to subsets of data
29. ditto for extinction coefficients
30. Molecular weight composition analysis (NNLS)
31. Open source/GPL/LGPL licensed, and free to download.
32. binary packages should be available for a wide range of platforms
33. ANSI C++ programming language used without proprietary/undocumented extensions
compilable by a wide range of compilers and systems
34. stochastic noise/autoconverge feature that allows overcoming of local minima in fit
35. linear constraints
36. ability to interface with Tom's new fluorescence database and operating system

Well, that about sums it up. As UltraScan users may have noticed,
I just described the features of the UltraScan equilibrium fitter. All
but the last 3 points are implemented. Some still need more work and some
features have some bugs that need attention (nothing is perfect), but all
in all very usable (especially by novices) and stable, and has been for
quite some time. Yes, Karin, OSX is available. 

having said all of this, I want to make a proposal to the software
developer community:

I would like to put the UltraScan source (all 150,000 lines of code,
not just the equilibrium stuff) on sourceforge and make it completely
open source as a benefit to the community, with the idea that we could
combine all of our collective creativity and each AUC software guru could
contribute their best ideas to a common product that will in the end
be the best service for the community of AUC users, novices and experts
alike. A community software project may in the long run be better than
everyone trying to reinvent the wheel. Let's reuse and improve what's
already out there.

For those interested, I already make source code available, although the
license is not yet GPL, but if enough serious interest and feedback exists
I will go through the trouble with my University's legal department to
push that through (they own the copyright on UltraScan rights according
to our rules).

So: If there is interest in this idea, let's take this idea offline
and discuss the details. I have some ideas on how it can be done.
Please mail me directly if you want to discuss this further.

Cheers, -Borries