XFF · parameter server
XtalPi Force Field (XFF)
XFF is an AMBER-compatible force field for drug-like molecules. It achieves outstanding performance in reproducing QM energies and geometries, as well as free energy perturbation (FEP) accuracy.
Already registered? Sign in
How it works
From molecule to parameter files
-
Provide a molecule
Upload a mol or sdf file (V2000 and V3000 are both supported), or a mol2 file that already carries its charges.
-
Add partial charges
Upload the charges as a separate file, or use a mol2 file that already carries them. Supported charge formats.
-
Download the parameters
Choose Amber, CHARMM or GROMACS and get the parameter and topology files in one archive.
Partial charges
Bring your own charges
The recommended partial charge model for XFF is RESP. You can obtain it with AmberTools and Gaussian, and upload it as a separate charge file.
mol and sdf files carry no partial charges, so they always need a charge file. Without one every charge in the output would be zero, and the server asks for the file instead of running.
Command line
Download GENXFF
Prefer to run it locally? GENXFF is also distributed as a command line tool. Write a small config file, point it at your molecule and charge files, and it writes the same parameter files.
- Linux x86 only
- Licence agreement required
By downloading GENXFF or getting XFF parameters online, you agree to our license agreement.
# generate parameters from a config file
$ genxff --config example.config
# dump the complete config template
$ genxff --dump_config template.config
Release history
Details for different versions of XFF
-
Fix the torsion parameters related to the following structures:
-
The version used in the XFF manuscript.
Note: this deployment runs a genxff build with the XFF 3.0.1.8 parameters compiled in, so 1.1 is the version it produces. Producing 1.0 needs its parameter file installed on the server.
References
-
Bai Xue, Qingyi Yang, Qiaochu Zhang, Xiao Wan, Dong Fang, Xiaolu Lin, Guangxu Sun, Gianpaolo Gobbo, Fenglei Cao, Alan M. Mathiowetz, Benjamin J. Burke, Robert A. Kumpf, Brajesh K. Rai, Geoffrey P. F. Wood, Frank C. Pickard IV, Junmei Wang, Peiyu Zhang, Jian Ma, Yide Alan Jiang, Shuhao Wen, Xinjun Hou, Junjie Zou, and Mingjun Yang. Journal of Chemical Theory and Computation 2024, 20 (2), 799–818.
Contact us: xff@xtalpi.com
About XtalPi
Visit xtalpi.comXtalPi (XtalPi Holdings Limited, 2228.HK) is an AI for Science company building infrastructure for autonomous scientific discovery. Grounded in quantum physics and powered by AI and robotics, XtalPi connects Agentic AI, Scientific AI, and Physical AI to bring scientific reasoning into the laboratory and create a continuous loop of prediction, experimentation, and learning.
By integrating scientific intelligence with large-scale robotic experimentation, XtalPi aims to turn AI into scalable scientific productivity—accelerating the discovery of innovative medicines, new-energy technologies, advanced materials, and chemicals, and shortening the path from scientific insight to real-world outcomes.