The quality of maps generated using 2D electron crystallography is limited owing to various factors including: (i) the problem of missing cone which occurs due to the fact that micrographs can be recorded until a certain tilt angle typically up to 60 degrees; (ii) the images of tilted samples have much lower quality than those of non-tilted ones; (iii) the current methods such as Lattice line fitting used to generate 3D data lack reliability. An iterative method applying known constraints in real and reciprocal space is devised in 2DX. These constraints exploit the fact that the densities mimic protein structures.

In 2DX we can refine the maps using the script “Refine Merged Map (3D only)”. There are several parameters in the section “3D-Volume Refinement” which should be properly set to get meaningful refined structure.

Expected membrane height: If the protein you work on is a membrane protein, then one can specify the height (in A). The densities will then be restricted to this height.

Low pass filter resolution for mask: In each iteration, the current density is first low-passed to create a blurry version of the current map. This is used as a mask to filter out noise and signal.

Higher and lower density thresholds: These thresholds suggest the density limits assumed for the noise. Everything below the lower threshold will be deleted. The densities which lie between higher and lower threshold are linearly scaled down.

Factor to scale the optimum energy of recovered cone: Due to masking and noise remove the amplitudes of the remaining reflections can go higher. This parameter can be used to scale the energy in each iteration.

Amplitude cutoff: The amplitude value below which all the reflections would be ignored. This can be changed to speed up the refinement.

After you run the script, there would be many files generated. Look at the masks and final volumes. Analyze if because of the masking the volumes being cut are reasonable. The aim should to remove as much noise as possible keeping as much signal as we can. Try different parameters if you think the masking is not reasonable.