difference between frealign 9 and frealix


I saw that frealign v9.x now also has more helical features implemented - is there much difference between the two programs besides presumably differnt coordinate systems?

Hi Matthias,

The main difference lies in the restraints on the alignment.

Frealign 9 has a filament stiffness parameter, which restrains alignment parameters (psi, theta) based on the assumption that filaments are almost straight & rigid. This works very well for many filaments, e.g. TMV, microtubules...

However, in cases where filaments have significant curvature (e.g. thin amyloid fibrils) this restraint is not appropriate.

Frealix implements spline-based restraints which take into account filament curvature. This results in better alignments in some cases. See for example Figure 5 (panels B and C) of 4402, where a set of curved filaments were first processed with Frealix and then refined further with Frealign, leading to deterioration of the map.

In addition, Frealix allows for restraints on all other parameters (including helical parameters) because it has a built-in model of every subunit in every filament, which is continuously updated and can be used to inform the alignment. Using all restraints in this way is computationally expensive, but may be helpful/informative in some projects.

Generally, Frealix is designed be the go-to solution where there is no good compromise segment length value that can be used in traditional single-particle-like processing strategies. In other words, for thoses cases where the SNR or filament thinness require long segments but the filament disorder requires short filaments and no single segment length can be found to accomodate both goals.

There are many other small differences, but these are the main ones.