The proposed procedure is a combination of tools available in the BioShell package [32] and commercial Sybyl software (Tripos Inc. St. Louis, MO) integrated in a single pipeline with the reduced-space CABS [17] protein folding algorithm, which can be employed in a high-throughput protein modeling.
Decoys generation by CABS
The CABS model uses a lattice representation with 800 possible orientations of the virtual alpha carbon-alpha carbon bonds [17]. The sampling scheme of the conformational space employs the Replica Exchange Monte Carlo method. Knowledge-based potentials of the force field include: generic protein-like conformational biases, statistical potentials for the short-range conformational propensities, a model of the main chain hydrogen bonds and context-dependent statistical potentials describing the side group interactions. The model could be effectively used for high resolution comparative modeling as well as for purely de novo folding of small globular proteins.
The Bioshell package was very useful in managing and analyzing large volume of simulation data.
Backbone reconstruction
Reconstruction procedure from alpha carbons to backbone, implemented in Sybyl/Biopolymer, with default settings was used. Procedure bases on a "spare parts" approach [26], using fragments retrieved from the protein database (PRODAT) to construct the full poly-alanine backbone. Subsequently side chains were added by the standard procedure implemented in Sybyl/Biopolymer with the initial side chain position from the Sybyl database.
All-atom minimization
We run all-atom minimization for all decoys (over 5500 structures from the CABS decoys' set and 2100 from the Moulder testing set) with frozen alpha carbons using implemented in Sybyl Amber7 ff99 force field, Amber charges, dielectric constant equal 1.0 and Powell minimization method, without initial optimization.
Amber7 ff99 energy expression:
Etotal = Estr+ Ebend + Etor + Evdw/ele, (Estr - Bond Stretching Energy Term, Ebend - Angle Bending Energy Term, Etors - Torsional Energy Term, Evdw/ele - van der Waals/Electrostatic Energy Term)
Maximum number of iterations was set to 5000, however the simulations longer than 1000 iterations were not necessary. To speed up maximally the calculations, a sufficient number of iterations (resulting in model ranking similar to the unlimited minimization) should be range of 500–1000, or even shorter, depending on the accuracy of models.
CABS and Bioshell package can be downloaded from our website [33].
Answers to all your Biology Questions
