<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Hi Mark,<br>
<br>
thanks a lot for your positive and instructive feedback - it is very
encouraging! Thank you!<br>
<br>
Yes, the automation of structure refinement is "a work in progress",
indeed. But I believe we are making a good progress with a positive
gradient! And the progress we have right now highly encourages us to
make further and further steps.<br>
<br>
I honestly and, perhaps, naively believe that the complete automation
in crystallographic structure refinement is possible and is subject of
time and effort invested into it. I clearly understand and foresee how
the next steps in automation look like and how they should be done, and
we keep going towards achieving this goal. One of the things where I
might fail is the time frame since all projected times for development
should be multiplied by pi or, better, by pi**2, to be realistic -:)<br>
<br>
For example, it took me a few weeks to work out the math and code the
basic bricks for TLS refinement, but then I spent nearly 6 months
making it rock solid reliable, robust and failure-proof (compared to
the other software options).<br>
<br>
Right now I'm working on real space refinement with the ultimate goal
to have a dual-space refinement, where the refinement in reciprocal
space (what we have right now) will be interleaved with the local
refinement in real space. Hopefully, this will be an efficient tool for
fixing local model imperfections that people used to do manually in
graphics (putting side chains into density, etc...). Of course, it
might be only possible with the power of torsion angle parameterization
that Ralf is currently working on and is about to finish.<br>
<br>
You are right, 100% correct handling of alternative conformations is
another big project that is in out to-do list. We did some progress in
this direction but had to switch to other pressing things. One day it
will be covered.<br>
<br>
And yes, sometime we need to write papers and make tutorial slides to
make it clear what we are busy with (which takes a lot of "development
time"), but it doesn't sound like a waste of time in the end! -:)<br>
<br>
Thanks again for your feedback,<br>
Pavel.<br>
<br>
<br>
On 3/8/09 7:43 PM, Mayer, Mark (NIH/NICHD) [E] wrote:
<blockquote
cite="mid:C19D5857C4C82743ACAFD281BA3169D50170D0F7@NIHCESMLBX7.nih.gov"
type="cite">
<pre wrap="">Hi Pavel,
These are excellent tools for students and the working crystallographer;
no need to apologize to sending to BB. For the 1st time I understood
value of automation in refinement. But, its still a work in progress, good
for everyone these days, and perhaps puts forward an ideal situation,
bit different from what some of us encounter in real life.
What would be great in future would be examples and discussions about
how/when to go anisotropic at resolutions where this is clearly warranted
but not for the the whole chain, (e.g. we have current structures @ 1.6 Å
which still have loops with almost no density) so advice on how to define
boundaries for isotropic and anisotropic refinement in single chain would
be great.
It was also interesting to see in your PDF that building still plays a key role in
structure solution; its obvious to anyone solving structures that
manual intervention is still needed here. Definition for refinement
of residues and solvent with coupled ACs still requires a lot of manual
intervention after building and automation before refinement would be
a real help!
Thanks again for all your contributions.
Mark Mayer
</pre>
<pre wrap="">
<hr size="4" width="90%">
_______________________________________________
phenixbb mailing list
<a class="moz-txt-link-abbreviated" href="mailto:phenixbb@phenix-online.org">phenixbb@phenix-online.org</a>
<a class="moz-txt-link-freetext" href="http://www.phenix-online.org/mailman/listinfo/phenixbb">http://www.phenix-online.org/mailman/listinfo/phenixbb</a>
</pre>
</blockquote>
</body>
</html>