Page 122 - Macromolecular Crystallography

MOLECULAR REPLACEMENT TECHNIQUES 111

Protocol 7.6 Full 6D search with the phased translation function

1. Determine the asymmetric unit of the space group of the The resulting formatted file will be fed to AMoRe as m1.rts,
rotation function of your space group (Urzhumtsev and the input file to Phased Translation Function (PTF).
Urzhumtseva, 2002). 3. Run AMoRe PTF in all the possible space groups (see
2. Write and execute jiffy code to create the (formatted) Protocol 7.1), after having run a test case (as in
input file to explore exhaustively rotation space: Protocol 7.2).
c... 4. Use perl to extract the information on the solutions with
delta0 = 5. the highest score from the (enormous) log file. Sort the
delta1 = delta0/2. solutions.
doi= −imax,imax 5. Plot results using gnuplot.
do j = 0,jmax
dok= −kmax, kmax
alpha = alpha0 + delta0*i
beta = beta0 + delta1*j
gamma = gamma0 + delta0*k
write(2,’(3f10.5,4x,2f11.3,i10)’)alpha,beta,gamma,r,t,iu
enddo
enddo
enddo
c...

the authors can keep track of both the successes normally using the best combination of weighted
and failures of jobs submitted by the crystallo- model(s) identified in this way.
graphic community and use these as bench marks
and opportunities to improve their protocols. So
Acknowledgments
their performances should keep growing, provided
that crystallographers don’t shy away from them We gratefully acknowledge financial support from
because of confidentiality problems. ACI-IMPB045 from the Ministere de la Recherche
Finally, by way of setting perspectives, we wish to et de la Technologie and from GDR 2417 du CNRS.
point out that the only alternative to the use of many We thank G. Labesse (CBS, CNRS, Montpellier)
different models successively, with the same proto- for discussions and communication of results prior
col, until one of them gives a better signal, would to publication and P. Dumas (IBMC, CNRS, Stras-
be to use all of them simultaneously, weighted by bourg) for comments on the manuscript.
some linear prefactor that remains to be determined.
The sum of all these prefactors for all the different References
copies of the model should of course be 1. This is
actually reminiscent of some recent approaches to Adams, P., Pannu, N. S., Read, R. J., and Brünger, A. T.
(1999) Extending the limits of molecular replace-
the refinement of macromolecules using a multicopy
ment through combined simulated annealing and
strategy, so as to take into account some inherent
maximum-likelihood refinement. Acta Crystallogr. D 55,
flexibility of the model (Burling and Brünger, 1994;
181–190.
Pellegrini et al., 1997). We suggest that refining the Altschul, S. F, Madden, T. L., Schaffer, A. A., Zhang, J.,
weights of the different models at each orientation Zhang, Z., Miller, W., and Lipman, D. J. (1997)
of the model in the rotation function should improve Gapped BLAST and PSI-BLAST: a new generation of
the signal-to-noise ratio of the whole procedure. protein database search programs. Nucleic Acids Res. 25,
Translation function searches should then proceed 3389–3402.