Re: model free analysis: spin deselection problem -- November 23, 2015 - 10:55

On 20 November 2015 at 14:17, Troels Emtekær Linnet
<tlinnet@xxxxxxxxxxxxx> wrote:
> Hi.
>
> My colleagues Kresten Lindorff-Larsen and Amelie Stein kindly helped me find
> some references:
>
> http://zhanglab.ccmb.med.umich.edu/HAAD/
> http://zhanglab.ccmb.med.umich.edu/papers/2009_7.pdf
>
> Fortran code:
> http://zhanglab.ccmb.med.umich.edu/HAAD/haad.f90
>
> http://kinemage.biochem.duke.edu/software/reduce.php
> http://kinemage.biochem.duke.edu/software/readme.reduce.html
> http://kinemage.biochem.duke.edu/php/downlode.php?filename=/downloads/PDFs/1999WordJMB285b.pdf
>
> I don't know the license of the programs, and if they should/could be
> implemented in relax efficiently by
> running the code as external program in the respectively software
> environments.

Hi Troels,

Thanks for digging out those references.  The HAAD algorithm and
software seems promising.  The online server could be used to create a
number of system tests (the check out all different molecule types),
and then the algorithm implemented to exactly match the HAAD software
results.  In the future, maybe we could ask them about dual licensing
a copy under the GPLv3+ licence to allow us to translate the FORTRAN
code directly into Python, as most of this seems to be hardcoded
numbers anyway.  I also cannot find a licence for HAAD, but the last
sentence of their abstract sounds like they intended their code to be
public domain:  "Both an executable and the source code of HAAD are
freely available at http://zhang.bioinformatics.ku.edu/HAAD";.  So
maybe they'll be ok with the dual licensing idea to allow a direct
translation and addition into the relax library (maybe in
/lib/structure/protons/).

For Reduce, we could also do the same, but calculating the results
offline and converting them to system tests.  Then in the
structure.attach_protons user function, we can have the algorithm
parameter which can be set to 'HAAD', 'Reduce', etc., citing the
papers in the user function docstring.  Anyway, I don't have time to
implement these for now, but these are lots of great ideas for future
additions.

Cheers,

Edward