Regarding the search engines, the following calculations can be carried out for the two types of biological molecules:

1. Nucleic acids (DNA and RNA):
   • Energy minimization.
   • Canonical molecular dynamics.
   Only unrestrained calculation mode is at present available for DNA and RNA systems.

2. Proteins and peptides:
   • Energy minimization.
   • Canonical molecular dynamics.
   • Replica exchange (REMD) and multiplexed replica exchange molecular dynamics (MREMD).
   Both unrestrained and restrained mode is enabled with the MREMD search engine on the server. Not all restrained simulations are enabled on the server but can be run with standalone UNRES. Disulfide bridges can be considered either in the static or variable manner (the latter in MD/MREMD only). See Input data description for more details.

Energy minimization usually starts from the respective experimental structure and its purpose is usually to get a coarse-grained structure free from overlaps. The tutorial showing how to prepare the data for energy minimization for proteins is available here, for a protein with disulfide briges here and that for nucleic acids is available here.

Canonical molecular dynamics can be started from the respective experimental structure or from an extended or random-generated structure (subject to the non-overlap condition). Energy minimization is carried out before the MD run. The purpose of MD runs is to study the time evolution of a system or to determine the residue-wise fluctuation profile. Given limited resources of the server, starts from extended/random structures should be limited to small systems. The tutorial showing how to prepare the data for molecular dynamics for proteins is available here and that for nucleic acids is available here.

The purpose of REMD/MREMD is in-depth conformational search at a range of temperature. With the UNRES server, each REMD/MREMD calculation is followed by post-processing with the Weighted Histogram Analysis Method (WHAM) and cluster analysis to provide a selected number of representative structures (5 by default) at the selected temperature (280 K by default). A tutorial showing how to prepare the data for a regular MREMD simulation is available here.

The following restraints can be used with REMD/MREMD (proteins only):

• Bioinformatics restraints:
  • Secondary structure prediction (from PSIPRED, hhpred, etc.). Two formats are available: the condensed one, where each residue is assigned a coil, "-", extended, "E", or helix, "H" state or in terms of probabilities of each of these states of a given residue, the input being the "raw" output file from PSIPRED or hhpred. See Input data description for more details.

  • Whole templates (e.g., the structures provided by the ColabFold of AlphaFold servers). A number of templates can be input, even quite different ones for ambiguous predictions. The restraints are the distance and local-geometry restraints, in the forms of the negative log of the sum of Gaussians centered at the respective geometric parameters from each of the templates.

• Experimental restraints:
  • SAXS/SANS data: the distance distribution. The P(r) (probability distribution) table has to be copied and pasted into the respective field. Tutorials are avaiable here and here.

  • NMR data: These restraints have to be provided in a file. Three formats: text, NMR-STAR and NEF are supported; the files from the PDB can be uploaded directly. See Input data description for more details. A new feature is the possibiliy of running NMR-restrained simulations in the time-averaged mode; the respective tutorial is here.

  • Chemical cross-link mass spectroscopy (XL-MS) data: the restraints can be input as flat-bottom-well restraints, statistical pseudopotentials or MD-determined pseudopotentials. See Input data description for more details.

These calculations enable the user to do the docking of a peptide or a protein ligand to a protein target. Multiple docking poses are generated at random and a MREMD simulation is then started from the multiple structures thus generated. The results are processed with WHAM and clustered and the 10 top docking poses are visualized. A tutorial is available here.

These calculations enable the user to determine the aggregation propensity of peptides. By default, 8 peptide molecules are placed in a periodic box with size corresponding to the 0.002 M concentration; these settings can be changed in the Advanced mode. A MREMD simulation is started from a random-generated initial structure and the results are processed with WHAM and cluster analysis. The top 5 structures are visualized. The temperature profiles of the aggregation degree, distribution of oligomers and the relative (to that of isolated residues) solvent accessible surface area shown in addition to those of regular calculations. A tutorial is available here.