By Vinzenz Unger and Anchi Cheng

Changes of the CTF across the image of a highly tilted crystal require a modification to the simple protocol for cross-correlation and lattice straightening to account for the splitting of spots at higher resolutions. In practice this is achieved by replacing MASKTRANA by TTMASK, which is a hybrid between MASKTRANA and TTBOX. TTMASK will correct the transform for the TTF (see TTBOX) before applying a mask. Almost all of the input parameters are identical to those from the simpler program components. Accordingly, only the genuinely new features will be discussed here.

As discussed above, the major effect of the TTF is the splitting of reflections that lie in a direction perpendicular to the tiltaxis. Consequently, tight masking of the transform would not include any useful information for split reflections and, hence, the transform needs first to be corrected for the effect of the TTF before attempting to calculate a highly averaged reference image. Furthermore, an increasing mask hole radius is used for the loose masking of reflections that lie perpendicular to the tilt axis to allow all the distortion information to be passed on to the cross-correlation step. Accordingly, one needs to set a minimum and maximum radius in pixel (set $radius=“lower upper”) instead of a constant value like in MASKTRANA. For the tightly filtered image both values will still be equal 1. However, for the loosely masked transform the minimum and maximum radius will depend on the size of the image, the amount of tilt and the resolution. These values need to be optimized by trial and error. Be aware that in contrast to MASKTRANA, TTMASK will overwrite the inputfile. Therefore, keep a separate copy of the input transform if it seems likely that more than one run is required. Alternatively, the required transforms can be recalculated from the relevant images.

Another difference to MASKTRANA is that one cannot use a separate “spotfile” if ITYPE=1 to specify which reflections one would like to be included in the mask. However, as for MASKTRANA, ITYPE=0 invokes the lattice generator and creates all possible spots within the limits given for H and K (IH/IK min max) and the specified transform radius RMAX. Yet, the same caution note given earlier (see MASKTRANA) applies - using all possible spots may not be advantageous.

Detailed diagnostics for the TTF correction will be printed out for the number of reflections specified by parameter NSP. Finally, be aware that errors in the estimate for the underfocus or the tilt geometry will result in an erroneous unbending of the lattice. This is a pronounced difference to MASKTRANA where the only problems arise from letting too much noise through the mask. However, since TTMASK corrects for the TTF, any wrong assignments of image parameter will result in a distorted reference, which in turn will negatively influence the determination of the distortion vectors used by CCUNBENDE.

This and the fact that errors in the tilt geometry will also be perpetuated in the relevant parameter for CCUNBENDE can result in a situation where the “corrected” image is significantly worse than the original. This emphasizes again (see TTBOX), that the higher amount of sophistication also requires a much higher certainty that the chosen input parameter are correct.