...sorry but the "quickfold2 didnt work that well
for me... ;(
resolve does deliver some main chain to get an idea
that i have one domain clearly there but different vs
of resolve e.g 2.06 and newest vs give very different
results... i dont care about side chains right now..
what is a not so good map, right good question...
well i meant 3-2.8 Å and well dont have any quantitavive
measure right now.. of course you can see indeed by
eye whats in the map... just takes some time, doing it
once is ok of course...
i guess indeed i'll just try different things...
(still interested to hear opinions though...)
cheers,
tommi
Quoting Anastassis Perrakis <[log in to unmask]>:
>
> On Jul 31, 2007, at 15:24, Tommi Kajander wrote:
>
> > Hi,
> > i would be interested in hearing about people's preferences on
> > programs
> > for doign auto-tracing of protein chains (with not so great maps),
>
> I do like ARP/wARP (objective opinion).
>
> > so far
> > my feeling has been nothing is at least much better than resolve in
> > doing
> > this. but i was wondering if people would care to share examples on
> > cases
> > where there was some difference to what you started with...
> > ..of course one can always complete and correct by hand but when
> > you are
> > doing this with phasing iteratively it would be interesting to hear
> > opinions..
>
> A bit more seriously now, my feeling is as follows:
>
> 1. If the automated program does not deliver more than 80-90% of the
> structure,
> all of it reliably in sequence and without out of register errors,
> I would personally go back and do it in O. Being a big fun of Coot, I
> still like O better
> for building form scratch - I guess it is most likely just habit though.
>
> 2. Even if I would do things by hand, having a resolve, arp/warp,
> bucaneer, textal
> model in parallel can be helpful as guidance. I would run all these,
> it takes
> less time to run them than think if they can be useful or not.
>
> 3. arp/warp (.. yes,yes) can deal with extremely bad maps if high
> resolution data is available.
> I have seen it doing things I could not do by hand. I have also seen
> it (more often ...)
> to fail to trace 2.5 A maps that it would only take me a day or two
> to trace completely.
> so, what is a 'not so great map' is not clear to me. an awful looking
> 1.1 A map with 70 deg. phase error,
> is not the same as an awful looking 3.2 A map with similar errors,
> and a MAD 3.2 A map can be
> easy to trace either by hand or automatically. And sorry for the
> shameless plug as usual, but arp/warp
> does work at low resolution. not as well as in high resolution, but
> we do get successes (>75%) with some
> real 3.0 a datasets .. although not often ... but its worth a try.
> And of course the Quick Fold (albe) module
> of ARP/wARP can also be useful and it takes for 500 residues less
> time to run than me writing this email ;-)
>
> Tassos
>
> > thanks,
> > tommi
> >
> >
> >
> > --
> > Tommi Kajander, Ph.D.
> > Macromolecular X-ray Crystallography
> > Research Program in Structural Biology and Biophysics
> > Institute of Biotechnology
> > PO box 65 (Street address: Viikinkaari 1, 4th floor)
> > University of Helsinki
> > FIN-00014 Helsinki, Finland
> > Tel. +358-9-191 58903
> > Fax +358-9-191 59940
>
>
>
--
Tommi Kajander, Ph.D.
Macromolecular X-ray Crystallography
Research Program in Structural Biology and Biophysics
Institute of Biotechnology
PO box 65 (Street address: Viikinkaari 1, 4th floor)
University of Helsinki
FIN-00014 Helsinki, Finland
Tel. +358-9-191 58903
Fax +358-9-191 59940
|