New Generation of General AMBER Force Field for Biomedical Research Molecular simulation plays an essential role in biochemical and biophysical research. Its major application is to decipher molecular interactions between small molecule ligands and biomolecules, especially protein receptors, so that highly potent agonists or antagonists can be discovered to enhance or eradicate target functions. Despite tremendous efforts spent on development, it is still very challenging to accurately predict protein-ligand binding. A key element to a successful prediction is the quality of practical molecular mechanics force field (MMFF). From the viewpoint of feasibility, the classical additive force field is in a unique position to offer computational efficiency while maintaining robustness for accurate and automated parametrization, which cannot be easily afforded by a polarizable force field. The other key factor to a successful prediction is the ability of the sampling strategy to effectively sample “hidden” events that are coupled with state transitions. The major goal of this project is to develop and test the 3rd generation of GAFF (GAFF3) to significantly improve the quality of the general- purpose AMBER force fields. GAFF3 will be critically evaluated in studying biomolecule-ligand interactions using a novel GPU-accelerated 𝜆 -dynamics based orthogonal space tempering (OST) algorithm. The advanced sampling technique will guarantee that our macromolecule-ligand binding free energy calculations is not complicated by existing sampling issues so that GAFF3 can be objectively evaluated. We will first develop GAFF3 utilizing ABCG2, a new physical charge model which has demonstrated its superior performance in large scale solvation free energy calculations; New force field parameterization techniques, such as applying ANI-1x potentials to fast detect “bad” torsional parameters, will be extensively applied in GAFF3 development. We will then critically evaluate the GAFF3 performance in studying biomolecule-ligand interactions using both pathway-based and endpoint free energy methods. The OST sampling method will be developed and implemented for this evaluation effort. Last, we plan to apply a variety of strategies to handle “difficult” molecules identified by us or our users. Those strategies will include fine atom typing and introduction of new functional forms. We believe that those efforts will allow GAFF3 to approach the performance limit an additive model could have. We will also expand the chemical space of GAFF3 to cover those elements not covered by the current GAFF, but frequently occurring in drugs and PDB ligands. Therefore, the successful pursuit of these research aims will facilitate us to surmount the challenges in accurately modeling protein-ligand and nucleic acid- ligand binding.