Major Changes in Freeform (freeform -calc conf)¶
- Tracking input conformers is no longer controlled by the -input3D flag (which has been removed). This is now done by specifying the file containing the conformers to track after the -track command line flag.
- Tracking multiple input conformers at once is now possible.
- Better solvation energies are now accessible as Poisson-Boltzmann single-point solvation energies at the Sheffield-solvation minima.
- Better conformational sampling in Freeform’s first stage.
New features in SZYBKI¶
- A SMARTS pattern is no longer required to constrain torsion with harmonic potential (option tor_constr). This can now be done by providing 4 atom indices that define the torsion.
- Options residue, sideC and polarH are now deprecated and have been replaced by a combination of the flextype option and flexdist or flexlist. The latter allows user selected residues to be become flexible during bound ligand optimization.
- Beginning with this release, the default value of the protein_elec option is ExactCoulomb. In all previous versions, no protein-ligand electrostatic interactions were used by default.
New features in freeform -calc conf¶
- The method for tracking input conformers has been changed. The input file containing these conformers must now be specified after the -track flag. The old method of using the flag -input3D or -useInput3D will not work and will cause a command line error.
- The command-line flag -input3D or -useInput3D has been removed; the equivalent functionality is now performed by specifying an input file with the -track flag.
- More than one input conformer per molecule can now be tracked in one run. The input conformers must either be successive conformers in an OEB file or successive structures of identical chemical graph, atom ordering, and bond ordering in another acceptable format.
- A new flag -solvent has been added, allowing users to control whether they want a Poisson-Boltzmann corrected Sheffield solvation or no solvent at all (i.e., gas phase) as alternatives to Sheffield solvation (the default).
- The new default method for applying partial charges to a ligand is AM1BCC ELF10, allowing up to 10 diverse conformers to be selected from conformers having the Electrostatically Least-interacting Functional groups (ELF). These conformers are then charged with the AM1BCCSym method and the charge sets averaged to come up with a single charge set applied to all conformers. This yields good-quality charge sets even for charged ligands. Previously, neutral versus charged ligands behaved differently: neutral ligands were charged using AM1BCCSym on a single ELF conformer, while charged ligands received MMFF94 charges for the sake of consistency.
- The initial conformational search is now performed using the full MMFF94S forcefield with Sheffield solvation, and with an energy window increased to 15 kcal/mol. This leads to improved performance particularly for molecules having internal hydrogen bonds.
- A new flag -ewindow has been added. This flag sets the energy window used to accept or reject conformers in the initial conformational search stage in a manner strictly analogous to the flag of the same name used in OMEGA or the SetEnergyWindow parameter in Omega TK.
- A new flag -maxconfs has been added. This flag sets the maximum number of conformers allowed in the initial conformational search stage in a manner strictly analogous to the flag of the same name used in OMEGA or the SetMaxConfs parameter in Omega TK.
- A new flag -solvent_dielectric has been added, allowing users to specify a dielectric other than 80.0 (the default) for the solvent dielectric of the Poisson-Boltzmann calculations in PB corrected solvation energies.
- A new flag -forcefield has been added, allowing users to specify the MMFF94 forcefield instead of MMFF94S (the default) for the freeform -calc conf energy minimizations. The conformational search still uses MMFF94S.
- To aid in analyzing the results of tracked conformers, three new output files have been added, with suffixes *.tracked_input.oeb, *.tracked_rstr.oeb, and *.tracked_free.oeb. These contain the structures and key output values (as SD tags) for the tracked conformers’ input structures, restrained-optimized minima, and freely-optimized minima, respectively.
- Zero-point energy is now included together with the internal energy and solvation energy when calculating the partition function. This quantity is calculated based on the same vibrational states used for the vibrational entropy.
- The output log file has changed its contents and format to make it easier to locate and extract key values.
- The output CSV file has several fields that contain additional values relevant to conformer free energies.
- A discontinuity has been removed in flat-bottom potentials (harm_constr1 followed by harm_constr2) that caused poor convergence during optimization and/or incorrect values of the harmonic potential to be reported in the log file.
- With the flextype option set with polarH the printed list now includes protein polar hydrogens that are flexible during bound ligand optimization. Previously, only a list of side chains or complete flexible residues was printed.
- An error message is now generated when trying to use Newton optimization when
- a protein-ligand electrostatic model is not ExactCoulomb or None,
- a free ligand is not in vacuum or other than Sheffield solvation model is used,
- a subset of atoms is fixed, or
- a harmonic constraint potential is applied.
- When AM1BCC charges are selected in Sheffield or PB solvation models for multiply charged species (with option -am1bcc), the SCF procedure in AM1 might not converge within 20 cycles. To make sure that the calculated atomic charges are stable, up to 1000 SCF iterations are allowed for multiply charged molecules.
- Option max_iter no longer can be set to a value of zero.
- SZYBKI output molecular files in OEB format contain a number of calculated numeric data. Previously, if such files had been used as input files in a subsequent SZYBKI run, the results reported in the log file could have been wrong. This problem has been fixed.
- In freeform -calc conf, using an existing input ensemble was previously restricted to molecules with 20 or fewer rotatable bonds. This limitation has been removed.