I'm not against the old style of coffee, I just don't know why anyone
would want to drink it now that the new stuff is available. I am not
against choice, it just needs to be clear to me why it is necessary.
Having thought about this a bit, I think that there are two major
differences in the real space refinement in Coot since 0.8.9.
i) The non-bonded contacts were modelled with an harmonic
approximation and now use Lennard-Jones. The harmonic code is still
in the source files but is not in the execution path. With a
judicious adjustment of a few if statements, it can be
re-established.
For now, you can adjust the LJ weights like this:
set_refinement_lennard_jones_epsilon(0.005)
That will soften up the NBC and may allow some interactions that
are repelled using the default weights.
Also (as I said) you can use the interactive contact dots to keep
a visual track of atom overlaps.
ii) The rigid-body "rubber-sheet" proportional neighbour drag has
gone. Initially I thought that it was unnecessary (and normally
that's the case), but I see that it can be useful and is now in
0.9.1-pre.
The introductory tutorials were written a long time ago and don't
mention key-bindings ("hot keys") or interface customisation. Perhaps
they should. Here is a more modern tutorial that mentions interface
customisation:
https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/web/tutorial/Coot-Cryo-EM-basics.html
Now that I read it, I see that it needs to be updated - Curlew is now a
full member of the Coot GUI (found under "File").
You can see me tweaking the interface in a video I did for SBGrid (it
was only meant for SBGrid attendees really):
https://www.youtube.com/watch?v=hhB8qUBBnJQ
Although note that the interface has changed (improved?) since then -
I need to redo that video.
Here is the ligand fitting/ribose rotation that I mentioned the other
day:
https://www.youtube.com/watch?v=HTNsBBXAc78
I need to redo that video too (it was a quicky I made for Frank), but
maybe better than nothing to illustrate the utility of the new style of
atom dragging.
The multi-threaded update to the refinement was tricky to write and
now it behaves pretty stably for me but there were quite an number of
brain-twisting timing/synchronisation issues that I needed to work
out. I admit that I get "Failed: Error No Progress" more frequently
than I did when I used 0.8.x. I wonder if that's what you're seeing. I
wonder if your hardware, your map and model, your mousing skills are
tickling a race condition that I don't see (which in the coffee
analogy would be "The light is on on this new machine you've given me,
I press the button, but nothing comes out"). But then again, no one
has mentioned a bug, merely an unfamiliar behaviour. So I don't know
if the software in your hands is working as intended and we are having
a discussion about model building workflows or, rather more
prosaically, what you're seeing is just a bug.
I had said:
many people learnt (I discovered, rather late in the day)...
What I meant by that was two things in particular: (i) not Ctrl
dragging (as we've discussed) and (ii) (it seemed to me) that people
were asking the RSR to do too much (i.e. starting too far from where
they wanted to end up). Quick Transformation (Ctrl Arrow keys), jiggle
fit and eigen-flip bring the residue or ligand close to the target
position before RSR is initiated. These are meat and drink for the
Coot tutorials I teach at courses.
Eike wrote:
No special key-binding, no scripts - just out of the box, default
settings, ... Eigen-flip, jiggle-fit, pepflip, JED-flip, backrub,
interactive contact dots, acedrg – sorry I’ve never heard of them.
Well if you add to that list pyrogen, lbg, flev, lidia, LO/Carb,
ligand validation, Curlew, the bespoke colour system, Geman-McClure,
RNA base and metal restraints you will be missing almost all of the
work that I've done on Coot since 2009... :-), :-( or maybe :-/
Relatedly:
Often we are restrained in time and improving your toolbox in an
unexpected manner is harder...
I recognise this and am trying to act on it - in the new (new) Coot - the
interface is quite a bit different, I am trying to use more familiar
idioms.
Herman said:
> The old real-space refinement was intuitive and easy to use and
did exactly what the user expected, without having to consult the
manual! The result might not have been perfect, but was good
enough for subsequent Refmac, Buster, Phenix refinement.
I am glad you think so. I thought so too. But I also think that 0.9.x
is even better at doing that.
Eike:
I also make use of the stepped-refine procedure
I no longer use the "All Molecule" menu items and considered removing
them from the interface. There is a function (fill_partial_residues)
to cover what the "Fill Partial Residues" menu item does - and does
them all at once. The improvement in refinement means that the length
of the molten zone can be increased (as you can see in the above
video). And this one (which Rob showed at the Study Weekend, highlighting
the value turning on GM restraints):
https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/files/movies/refine-example-nov.ogv
I also make use of the stepped-refine procedure, but I know more
than one colleague who considers that “cheating” already
Well, for the record then I am OK with using stepped refine for
filling in missing residues, but using the Refine/Improve Rama
automated and blind (i.e. you go for a coffee while it runs (if you
can find a working coffee machine)) - that I find problematic too.
Robbie:
3) Practice. The C-termini in PDB entry 1ggx are nice to practice
resisting to over-drag.
Additionally, the Madrid version of the "Wonky N-terminus"" demo box is
available from the web site.
https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/files/demo-madrid.tar.gz
The Ferrari thing was a reference to presentations by Alywn Jones - he
used to talk about computers for graphics costing n Ferraris (n>>1 IIRC).
So, as a result of this discussion, I will prioritise 0.9.1 and get it
out the door so that we can continue this discussion next week (or so) in
the light of proportional editing.
The Coot mailing list would have been a more appropriate forum than this
one. We don't do top-posting there.
I still remain very curious about what you find easier in 0.8.x
compared to 0.9.
Paul
On 09/09/2020 16:22, Georg Zocher wrote:
Dear Paul,
people are complaining that they have issues fitting their model in
real space using coot 0.9.x. Personally, I very briefly used coot
other the last months but also have problems using RSR properly
because I'm used to do it the way it was over the last 10 years. This
might be due to my ignorance that I did not have a more closer look
into the new features of RSR in 0.9 release and how to use them. But
from the user site it's different to handle now.
I do not understand your Ferrari comparison (sorry) as I would not
buy one even if I would have more money than one can spend. But I
would take a cup of coffee as an example. If you always go to the
same dealer to get your coffee in the morning as you know you get a
very descent and excellent tasting cup of coffee that you fully enjoy
every morning than you really get used to it. It might be than a hard
time for you if your coffee dealer tells you: Sorry guys, I do have
now an optimized coffee that is much better, taste a lot better, is
fair produced,... It's just not the coffee you drunk over the last 10
years and I will not offer that one anymore. You might give the new
one a chance, you might find it as excellent as the dealer but you
might not. From my point of view it's the lack of choice that
personally I do not like so much, especially if there is no other
coffee dealer around...
Nevertheless, I'm aware and fully respect all the effort you put in
the development of coot and I'm really grateful that coot is available.
All the best,
Georg
Am 08.09.20 um 17:44 schrieb Paul Emsley:
On 08/09/2020 16:25, Georg Zocher wrote:
we have the same experience in our lab.
What experience is that? I am still in the dark about you think is
now worse.
Personally, I did would not like to judge here, as so far, I did
not have had enough time to get into the new RSR of coot 0.9.x by
myself. But many colleagues did not like the new refinement module
maybe just as they are used to the method in all coot versions before.
You have a Ferrari parked beside your house but you want to to take
the bus to work because that's what you've always done. Or maybe the
Ferrari is parked around the back and you don't know it's there?
I just thought if it wouldn't be an option to let the user decide
what kind of RSR implementation she/he would like to use and give
them the choice via an option in coot preferences?
That would be possible but not easy. Unlike much of the CCP4 suite,
Coot is Free Software. But, again... why would you want to take the
bus? Explain.
regards,
Paul.
########################################################################
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a
mailing list hosted by www.jiscmail.ac.uk, terms & conditions are
available at https://www.jiscmail.ac.uk/policyandsecurity/
########################################################################
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a
mailing list hosted by www.jiscmail.ac.uk, terms & conditions are
available at https://www.jiscmail.ac.uk/policyandsecurity/
