An automated data processing and analysis pipeline for transmembrane proteins in detergent solutions

D. Molodenskiy; H. Mertens; D. Svergun

The electron densities can be either provided directly via the ISPyB interface or calculated from the protein primary sequence (in FASTA format) or from the chemical formula using the built-in look-up tables (37, 38) . Knowledge of the detergent chemical formula is mandatory as it provides additional information to constrain the geometry of the MP model. When the high resolution models are not available, ab initio low resolution bead modeling is performed by MONSA (27), allowing for three phases with distinct contrasts corresponding to the protein (phase 1), detergent tails (phase 2) and detergent heads (phase 3). For each phase the contrast is calculated and the search space confined to a cylinder and two coaxial spherical hemi-tori ( Figure 5A ), thus avoiding contacts between the hydrophobic regions and solvent in the final model. A spherical region is fixed for the protein phase in the vicinity of the origin of coordinates, while the rest of the cylindrical region is free to become either protein or solvent. Penalties for the model discontinuity and looseness are applied during the simulated annealing procedure as described in the original paper (27) . Three additional boundary regions between the phases ( Figure 5A ) have a thickness of a few Angstrom allowing MONSA to select the optimum phase assignment during the fitting. The boundary beads (eg. between phases 1 and 2) are allowed to be assigned to either phase 1 or phase 2 ( Figure 5A ). The model is sought within a spherical search volume of diameter d = D max , where the protein can be surrounded by twostacked tori (detergent tails and heads) with the lengths evaluated from the Tanford formula ( Figure 5B ). The transmembrane diameter is calculated as: