There have been a number of papers in the literature recently describing how to effectively "detwin" X-ray data collected from merohedrally twinned crystals (see Yeates, 1997, for a recent review). One of us (H.O.T.) has recently being attempting to solve a protein-DNA complex where the data show the classic signs of merohedral twinning (in the cumulative intensity distribution (N(z) test) tabulated in TRUNCATE). However, no program was available in the CCP4 suite to allow detwinning of the data.
A new program DETWIN has been written, which takes as input the MTZ file contained merged intensities written by SCALA, and writes out a new MTZ file, in the same format, in which the data have been detwinned. This file is then input to TRUNCATE in the usual way.
The detwinning is formally only possible for a twin fraction other than 0.5, and errors in the twin fraction will result in large errors in the detwinned intensities for twin fractions above 0.45. The twin fraction can be estimated from the N(z) statistics for the original data, or from DATAMAN (due to Gerard Kleywegt), or by running the program several times with different twin fractions and choosing the twin fraction that gives the best N(z) distribution or the smallest correlation between the intensities of reflections related by the twin operator (after an idea of Richard Henderson).
The program is very simple to run, but is not yet in its final form. In particular, it does not correct data with an anomalous signal. However, anyone wishing to try using it should send an E-mail to:
Once the code is in a more satisfactory form it will be released as part of the CCP4 package.
A more complete description of experiences using the program will be given in a later Newsletter.
Yeates, T.O. in Methods in Enzymology, Vol 276, 344-358, 1997.