AMBER - one of the most widely licensed molecular modeling and simulation packages in the world - provides high powered facilities for a large number of new simulation-based methods. Highlights of AMBER 6 include many new and useful features, as detailed below. It is our belief that these new features add great value to AMBER and fully justify the introduction of a new version of the package. We hope you agree and choose to license AMBER 6. Note that funds from licensing AMBER 6 support research by the authors of the package, and thus your fees support further development of new features.
The academic price for licensing is the same as for Amber 6, $400. (This fee may be reduced or waived in special circumstances.) The industrial price is $20,000 for new licensees and $15,000 for those who have licensed Amber 6.
To proceed with ordering AMBER 6, go to the
AMBER web site.
This site has much additional information, including the manual and
tutorials. To order AMBER, you can download the license agreement from
the web site, sign it, and send with an appropriate check or money order
made out to the Regents of the University of California, to
Willa Crowell.
AMBER 6 has a number of very exciting capabilities which should prove very powerful in structure-based ligand design and in the understanding of structure and free energy in any complex molecular system. New methods such as Chemical Monte Carlo/Molecular Dynamics (CMC/MD) allow the consideration of many molecules simultaneously and are a powerful complement to more qualitative approaches in drug design. One can imagine using DOCKing methods to screen databases of ligands and, after reducing the number of possible leads to leads ~100, using this method to more accurately estimate the binding free energy for these 100 compounds. Another major development has been the use of continuum methods combined with explicit MD and molecular mechanics. The method of computational Alanine scanning by Massova and Kollman (JACS, 121:8133, 1999) reflects the exciting possibility of modeling many protein mutants at once. In the era of genomics, use of MM-PBSA (Molecular mechanics- Poisson Bolzmann Surface Area) using molecular dynamics should lead to more accurate protein models and, when combined with homology modeled protein structures, should aid computer assisted design of tight binding ligands by making the protein structure more reliable and robust. The methodology of locally enhanced sampling (LES) has already been shown to be very powerful in protein structure refinement (Simmerling et al, JACS, submitted). New dynamics simulation methods to aid in ligand design(OWFEG and PROFEC) are available in AMBER 6. As additional bonuses, the excellent software for NMR refinement, the latest in force field models and parameters and the capabilities to model enzyme mechanisms (Stanton et al, JACS, 120:3448, 1998) are also part of the code. Below are some more technical details and references.
An overview of some of the new capabilities of AMBER 6:
AMBER 5 contained the capability for Locally Enhanced Sampling (LES) and Particle Mesh Ewald (PME), but not both together. These have been integrated and used in promising ways in prediction of RNA structure (Simmerling et al, JACS, 120:7149, 1998) and protein structure (Simmerling et al, JACS,submitted) as well as in free energy calculations of carbohydrate systems (Simmerling et al, JACS, 120:5771, 1998, and work in progress). We anticipate this methodology will be very powerful in structure prediction of complex molecules in solution, since, in contrast to most other methods, explicit water molecules can be included and thus the solvation effect considered at the highest level of accuracy.
Chemical Monte Carlo/Molecular Dynamics(CMC/MD) is a new approach to ligand design, which can consider many molecules at once and is a powerful complement to more qualitative approaches. In papers published by Pitera and Kollman (JACS, 120:7557, 1998) and by Eriksson et al. (J. Med. Chem., 42:868, 1999), it has been shown to be accurate and leads in both cases to a ligand which was predicted to bind more strongly to the target than the previous best one considered experimentally.
The OWFEG(One Window Free Energy Grid) method has been incorporated into the SANDER module of AMBER 6. OWFEG(Pearlman, J. Med. Chem., 42, 4313, 1999) allows one to generate an approximate free energy grid about any ligand from a single molecular dynamics trajectory. OWFEG is based on the PROFEC approach(Radmer and Kollman, J. Comput-Aided Mol. Des., 12,215, 1998), but has been generalized to be suitable for use with flexible ligands. As demonstrated in the OWFEG and PROFEC publications, such grids can provide very useful guidance in drug design.
Case et al. (JACS 120:9401, 1998) have shown that a combination of molecular dynamics simulations and continuum calculations can be very powerful in estimating free energy differences even between systems on which one cannot use free energy perturbation calculations. This methodology is automated in AMBER 6. This method has proven to be very accurate in calculations of protein-ligand interactions (Chong et al, PNAS, in press), in protein-peptide interactions (Massova and Kollman, JACS, 121:8133, 1999), RNA-protein interactions (Reyes and Kollman, JMB, in press) and protein folding (Lee et al, Proteins, submitted.)
Generalized Born surface area(GB/SA) is a way to put solvation effects implicitly into calculations of complex systems and recently Tsui and Case have incorporated this into AMBER 6 for use in minimization and molecular dynamics calculations. This approach allows one to greatly reduce the number of atoms in the system by representing solvent implicitly rather than explicitly. It is estimated that this saves more than an order of magnitude in simulations of proteins in aqueous solution.
AMBER 6 contains a major re-write of the particle-mesh-Ewald (PME) implementation for molecular dynamics in SANDER. This now accurately This now supports alternative box shapes(such as the truncated octahedron), allows polarizable potentials to be used in conjunction with PME and accurately conserves energy (in the NVE ensemble) over long trajectories. The user interface for PME calculations has been greatly simplified, so that in most cases the default parameters should give efficient yet acceptably accurate results. A variety of accuracy checks and comparisons to "regular" Ewald summation results are available.
NMR refinements can be carried out with restraints derived from residual dipolar coupling measurements or with "ambiguous" restraints whose corresponding NMR spectra are not fully assigned, or for "multiple-conformer" models generated using the LES algorithm. Routines to generate restraint input and to interface to NMR data-processing programs have been considerably expanded.
The ROAR 2.0 module contains QM/MM capabilities to carry out QM/MM minimizations, MD simulations and reaction path following studies. Semiempirical (PM3, AM1 and MNDO) theory is used for the QM part of the calculations, while the AMBER force field is the MM part of the calculation. Minimization can be done using steepest descent, conjugate gradient or through the use of a limited memory BFGS approach. The latter approach can also be used in fully classical minimizations.
The ROAR module contains the capability to carry out MTS simulations using the methodology of Berne and co-workers. Conservatively this allows a doubling of the time-step used, while is special cases much longer time-steps (e.g., 5fs) can be used. This option is also coupled with the Nose-Hoover Chain approach to constant T and P simulations and is designed to work with either standard Ewald or with PME. This option is only set up to run on condensed phases at this time.
The term "Amber" is also sometimes used to refer to the empirical force fields that are implemented in the code. It should be recognized however, that the code and force field are separate: several other computer packages have implemented the Amber force fields, and other force fields can be implemented with the Amber programs. Further, the force fields are in the public domain, whereas the codes are distributed under a license agreement.
Peter Kollman, who had inspired and led Amber development for more than two decades, died unexpectedly in May 2001. But many of the items cited below (particularly in force field development) were very near completion at the time of his death, and represent the culmination of several years of effort > from both Peter and his collaborators. The Amber development team is committed to continuing to update and improve the software and force fields, and we dedicate Amber 7 to Peter's memory.
Amber 7 (2002) represents a significant change from the current version, Amber 6, which was released in December, 1999. Briefly, the major differences include:
Several new force fields are available for proteins and nucleic acids. These include versions with polarizable dipoles on atoms, and off-center charges (also called "extra points", and analogous to lone pairs). Amber now provides direct support for TIP3P, TIP4P, TIP5P, SPC/E and POL3 models of water, as well as providing models for chloroform and other organic solvents.
A new "general amber force field" that should be applicable to most organic molecules. The automated code to prepare Amber input files using this force field is a new module, called Antechamber. In most cases, Antechamber can directly convert three-dimensional models into files appropriate for molecular mechanics calculations, automatically assigning atom types, charges and force field parameters.
Implementation of three new variants of the generalized Born (GB) code, including one that appears to provide a better energy balance between surface-exposed and buried atoms.
More efficient PME simulations, with better performance on both single-processor and parallel machines.
Updated scripts for MM_PBSA analysis, making input easier to create and providing more options for analysis of the results.
Free energy calculations using the thermodynamic integration method can now be carried out in sander. Many investigations that used to require gibbs can now be carried out in a simpler fashion, and free energy studies using the GB model or "extra points" force fields (which are not supported with gibbs) can now be undertaken.
New types of restraint forces can be defined that are based on RMS superpositions to reference structures. This "targeted MD" capability can be used to enhance or guide conformational sampling.
AMBER authors:
David A. Case, David A. Pearlman, James W. Caldwell, Thomas E. Cheatham III, Junmei Wang, Wilson S. Ross, Carlos Simmerling, Tom Darden, Kenneth M. Merz, Robert V. Stanton, Ailan Cheng, James J. Vincent, Mike Crowley, Vickie Tsui, Holger Gohlke, Randall Radmer, Yong Duan, Jed Pitera, Irina Massova, George L. Seibel, U. Chandra Singh, Paul Weiner, and Peter A. Kollman.
There is now an Amber mail reflector:
To join or unjoin:
amber-request@cgl.ucsf.edu
To post, mail to:
amber@cgl.ucsf.edu