Why not?
Bead modelling (see Byron, Hydrodynamic bead modeling of biological macromolecules, Methods Enzymol. 321, 278-304, 2000) has been successfully applied to a variety of different experimental systems. While even poor molecular representations (where models are only represented by a few beads) are useful to gauge the solution properties of a molecule, the benefit of a more accurate description is that the calculated properties are likely to be closer to experimental values. The only limitation with either approach is the availability of information for generating the starting model. It used to be the case that high resolution models were much scarcer for large macromolecules than electron microscopy images. However this has changed with advances in molecular techniques and the provision of automated procedures for screening protein crystals (Abola et al., Nat Struct Biol. 7, 973-977, 2000). The growth in structural information derived from experimental data and in genome sequences has led to the development of a wide array of structure prediction algorithms (Jones, Curr. Opin. Struct. Biol. 10, 371-379, 2000), making the development of computational tools for generating medium resolution bead models from high resolution atomic structures more attractive. Generating bead models from atomic data is essentially a process of reducing the information needed to describe the position of each atom in a macromolecule. The emphasis for hydrodynamic modelling is different from Debye sphere models. In order to conserve the overall shape of the hydrodynamic model, each bead is placed so that it reproduces the topological surface of the molecule. Whereas Debye sphere models attempt to reproduce the overall volume of the molecule and the mass distribution inside it. To do this, the volume of each bead is adjusted (with or without the inclusion of bound water) so that it reproduces the collective volume of all the atoms that it represents. Each bead is placed at the centre of gravity of a group of atoms that constitutes a scattering element. Most bead algorithms that utilise atomic data to generate the models work by grouping atoms into a single bead. This does not apply for shell models, where instead a layer of small beads is placed on the outside of the macromolecule, so that the surface of the model could compare favourably with the surface rugosity of the molecule. For hydrodynamic calculations the interior of a shell model is left void because it does not contribute to the friction. In order to perform SAS calculations, the model can be filled with equivalent sized beads so that the mass distribution from individual scattering elements is also represented. This is the basis for the HYDROPRO algorithm (Garcia de la Torre et al., Biophys. J. 78, 719-730, 2000), which generates both shell and 'shell filled' models. While the advantage of shell models is a detailed surface description, the number of beads that are included in the hydrodynamic calculations may compromise its speed. Instead SOMO, using the bead processing operators that will be described (see Rai et al., Structure 13, 723-734, 2005), can be used to generate medium (about a factor of four) resolution models. The bead processing operators in SOMO can be used in a limited number of arrangements to generate a variety of different models. The algorithms that are currently working are based on grouping many atoms into a single bead. In the case of the TRANS routine, each bead represents either a side or a main chain (and a single terminal bead that is treated by the processing operators as if it was a main chain bead). The AtoB (Byron, Biophys. J. 72, 408-415) algorithm groups atoms according to size of the grid placed over the atomic model.