In the referenced article, ohba at YL05HP.TB.NODA.SUT.AC.JP (Hiroyoshi Ohba) writes:
> Is there any software which assembles three-dimesional structure of
>antibody Fab? Now, I am not familier with structural biology but like to
>analyse the effect of molecular modification of antibody. I have tried AbM
>of Oxford Molecular Ltd which has been developed by Prof. Tony Rees's Lab.
>It works on the stand-alone Silicon Graphics and HP workstations. In my
>case on assembling human Fab from its amino-acid sequence, enormous time
>(days) was required for caliculation. Would you suggest quicker softwares
>on faster machines, or software employing other argorisms?
I would suggest playing with some of the settings within AbM first.
For any CDR loop which is not being built as a canonical loop,
model it as a database loop (from the BUILD screen, select the
BUILD METHOD for each CDR and enter "2" to select Database only
This will make a much quicker model, which will often be reasonably
good. I would suggest you make models using several different
methods for the CDRs, perhaps using different structures for
the heavy and light chain framework regions. Once you have a
set of models, compare them with crystal structures for similar
antibodies. If you superimpose your model with several
crystal structures where the CDRs are similar in length and
sequence, if the crystal structures are very similar to each
other, but different to your model, it is likely that the model
model is wrong.
The run times for models varies between antibodies, but on an
HP 735, we find that modelling the CDRs as canonical loops or
with database searches, they take between 5 min and 2 hours.
CONGEN and Database/CONGEN searches take longer, but not much
more than a day for most of the 30-40 antibodies that we have tried.
I am quoting these times from memory, but they should be correct
within a factor of 2 or 3.
School of Biology and Biochemistry
University of Bath
Avon, BA2 7AY
Tel (0225) 826826 ext 5411
Fax (0225) 826449
email A.H.Henry at bath.ac.uk