Guangfeng Zhou
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Guangfeng Zhou has a total of 90 co-authors. Here are the top co-authors by number of shared publications:
- Frank DiMaio
Institute for Protein Design, University of Washington, Seattle, WA, USA
10 shared publications - Vincent A. Voelz
Department of Chemistry, Temple University , Philadelphia, Pennsylvania 19122
10 shared publications - Hahnbeom Park
Korea Institute of Science and Technology
3 shared publications - Heike Wulff
Department of Pharmacology School of Medicine University of California Davis California USA
3 shared publications - Vladimir Yarov-Yarovoy
Department of Physiology and Membrane Biology, University of California Davis, School of Medicine
3 shared publications - Andrew Muenks
Department of Biochemistry, University of Washington, Seattle, WA 98195, USA; Institute for Protein Design, University of Washington, Seattle, WA 98195, USA
2 shared publications
Latest Publications
- Open Access
- 4 November 2025
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 122 (45)
Modeling protein–small molecule conformational ensembles with PLACER
Modeling the conformational heterogeneity of protein–small molecule interactions is important for understanding natural systems and evaluating designed systems but remains an outstanding challenge. We reasoned that while residue-level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state, an entirely atomic-level description could have advantages in speed and generality. We developed a graph neural network called PLACER (protein-ligand atomistic conformational ensemble resolver) trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. PLACER accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding. When given a description of the larger protein context, it builds up structures of small molecules and protein side chains for protein–small molecule docking. Because PLACER is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using PLACER to assess the accuracy and preorganization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with a k cat / K M of 11,000 M −1 min −1 , considerably higher than any pre–deep learning design for this reaction. We anticipate that PLACER will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule–protein systems and for designing higher activity preorganized enzymes.
- 4 November 2025
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 122 (45)
Open AccessModeling protein–small molecule conformational ensembles with PLACER
Modeling the conformational heterogeneity of protein–small molecule interactions is important for understanding natural systems and evaluating designed systems but remains an outstanding challenge. We reasoned that while residue-level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state, an entirely atomic-level description could have advantages in speed and generality. We developed a graph neural network called PLACER (protein-ligand atomistic conformational ensemble resolver) trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. PLACER accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding. When given a description of the larger protein context, it builds up structures of small molecules and protein side chains for protein–small molecule docking. Because PLACER is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using PLACER to assess the accuracy and preorganization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with a k cat / K M of 11,000 M −1 min −1 , considerably higher than any pre–deep learning design for this reaction. We anticipate that PLACER will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule–protein systems and for designing higher activity preorganized enzymes.
- Top Cited PapersOpen Access
- 21 May 2025
- Published by Springer Nature in Nature Communications
- Vol. 16 (1) , 4730
Cyclic peptide structure prediction and design using AlphaFold2
Small cyclic peptides have gained significant traction as a therapeutic modality; however, the development of deep learning methods for accurately designing such peptides has been slow, mostly due to the lack of sufficiently large training sets. Here, we introduce AfCycDesign, a deep learning approach for accurate structure prediction, sequence redesign, and de novo hallucination of cyclic peptides. Using AfCycDesign, we identified over 10,000 structurally-diverse designs predicted to fold into the designed structures with high confidence. X-ray crystal structures for eight tested de novo designed sequences match very closely with the design models (RMSD < 1.0 Å), highlighting the atomic level accuracy in our approach. Further, we used the set of hallucinated peptides as starting scaffolds to design binders with nanomolar IC50 against MDM2 and Keap1. The computational methods and scaffolds developed here provide the basis for the custom design of peptides for diverse protein targets and therapeutic applications.
- 21 May 2025
- Published by Springer Nature in Nature Communications
- Vol. 16 (1) , 4730
Top Cited PapersOpen AccessCyclic peptide structure prediction and design using AlphaFold2
Small cyclic peptides have gained significant traction as a therapeutic modality; however, the development of deep learning methods for accurately designing such peptides has been slow, mostly due to the lack of sufficiently large training sets. Here, we introduce AfCycDesign, a deep learning approach for accurate structure prediction, sequence redesign, and de novo hallucination of cyclic peptides. Using AfCycDesign, we identified over 10,000 structurally-diverse designs predicted to fold into the designed structures with high confidence. X-ray crystal structures for eight tested de novo designed sequences match very closely with the design models (RMSD < 1.0 Å), highlighting the atomic level accuracy in our approach. Further, we used the set of hallucinated peptides as starting scaffolds to design binders with nanomolar IC50 against MDM2 and Keap1. The computational methods and scaffolds developed here provide the basis for the custom design of peptides for diverse protein targets and therapeutic applications.
- Open Access
Automated identification of small molecules in cryo-electron microscopy data with density- and energy-guided evaluation
SUMMARY: Methodological improvements in cryo-electron microscopy (cryoEM) have made it a useful tool in ligand-bound structure determination for biology and drug design. However, determining the conformation and identity of bound ligands is still challenging at the resolutions typical for cry-oEM. Automated methods can aid in ligand conformational modeling, but current ligand identification tools — developed for X-ray crystallography data — perform poorly at resolutions common for cryoEM. Here, we present EMERALD-ID, a method capable of docking and evaluating small molecule conformations for ligand identification. EMERALD-ID identifies 43% of common ligands exactly and identifies closely related ligands in 66% of cases. We then use this tool to discover possible ligand identification errors, as well as previously unidentified ligands. Furthermore, we show EMERALD-ID is capable of identifying ligands from custom ligand libraries of various small molecule types, including human metabolites and drug fragments. Our method provides a valuable addition to cryoEM modeling tools to improve small molecule model accuracy and quality.
Open AccessAutomated identification of small molecules in cryo-electron microscopy data with density- and energy-guided evaluation
SUMMARY: Methodological improvements in cryo-electron microscopy (cryoEM) have made it a useful tool in ligand-bound structure determination for biology and drug design. However, determining the conformation and identity of bound ligands is still challenging at the resolutions typical for cry-oEM. Automated methods can aid in ligand conformational modeling, but current ligand identification tools — developed for X-ray crystallography data — perform poorly at resolutions common for cryoEM. Here, we present EMERALD-ID, a method capable of docking and evaluating small molecule conformations for ligand identification. EMERALD-ID identifies 43% of common ligands exactly and identifies closely related ligands in 66% of cases. We then use this tool to discover possible ligand identification errors, as well as previously unidentified ligands. Furthermore, we show EMERALD-ID is capable of identifying ligands from custom ligand libraries of various small molecule types, including human metabolites and drug fragments. Our method provides a valuable addition to cryoEM modeling tools to improve small molecule model accuracy and quality.
- Open Access
Modeling protein-small molecule conformational ensembles with PLACER
Modeling the conformational heterogeneity of protein-small molecule interactions is important for understanding natural systems and evaluating designed systems, but remains an outstanding challenge. We reasoned that while residue level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state an entirely atomic level description could have advantages in speed and generality. We developed a graph neural network called PLACER (Protein-Ligand Atomistic Conformational Ensemble Resolver) trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. PLACER accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding, and given a description of the larger protein context, builds up structures of small molecules and protein side chains for protein-small molecule docking. Because PLACER is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using PLACER to assess the accuracy and pre-organization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with a k cat / K M of 11000 M -1 min -1 , considerably higher than any pre-deep learning design for this reaction. We anticipate that PLACER will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule-protein systems, and for designing higher activity preorganized enzymes.
Open AccessModeling protein-small molecule conformational ensembles with PLACER
Modeling the conformational heterogeneity of protein-small molecule interactions is important for understanding natural systems and evaluating designed systems, but remains an outstanding challenge. We reasoned that while residue level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state an entirely atomic level description could have advantages in speed and generality. We developed a graph neural network called PLACER (Protein-Ligand Atomistic Conformational Ensemble Resolver) trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. PLACER accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding, and given a description of the larger protein context, builds up structures of small molecules and protein side chains for protein-small molecule docking. Because PLACER is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using PLACER to assess the accuracy and pre-organization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with a k cat / K M of 11000 M -1 min -1 , considerably higher than any pre-deep learning design for this reaction. We anticipate that PLACER will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule-protein systems, and for designing higher activity preorganized enzymes.
- Top Cited PapersOpen Access
- 5 September 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-14
An artificial intelligence accelerated virtual screening platform for drug discovery
Structure-based virtual screening is a key tool in early drug discovery, with growing interest in the screening of multi-billion chemical compound libraries. However, the success of virtual screening crucially depends on the accuracy of the binding pose and binding affinity predicted by computational docking. Here we develop a highly accurate structure-based virtual screen method, RosettaVS, for predicting docking poses and binding affinities. Our approach outperforms other state-of-the-art methods on a wide range of benchmarks, partially due to our ability to model receptor flexibility. We incorporate this into a new open-source artificial intelligence accelerated virtual screening platform for drug discovery. Using this platform, we screen multi-billion compound libraries against two unrelated targets, a ubiquitin ligase target KLHDC2 and the human voltage-gated sodium channel NaV1.7. For both targets, we discover hit compounds, including seven hits (14% hit rate) to KLHDC2 and four hits (44% hit rate) to NaV1.7, all with single digit micromolar binding affinities. Screening in both cases is completed in less than seven days. Finally, a high resolution X-ray crystallographic structure validates the predicted docking pose for the KLHDC2 ligand complex, demonstrating the effectiveness of our method in lead discovery.
- 5 September 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-14
Top Cited PapersOpen AccessAn artificial intelligence accelerated virtual screening platform for drug discovery
Structure-based virtual screening is a key tool in early drug discovery, with growing interest in the screening of multi-billion chemical compound libraries. However, the success of virtual screening crucially depends on the accuracy of the binding pose and binding affinity predicted by computational docking. Here we develop a highly accurate structure-based virtual screen method, RosettaVS, for predicting docking poses and binding affinities. Our approach outperforms other state-of-the-art methods on a wide range of benchmarks, partially due to our ability to model receptor flexibility. We incorporate this into a new open-source artificial intelligence accelerated virtual screening platform for drug discovery. Using this platform, we screen multi-billion compound libraries against two unrelated targets, a ubiquitin ligase target KLHDC2 and the human voltage-gated sodium channel NaV1.7. For both targets, we discover hit compounds, including seven hits (14% hit rate) to KLHDC2 and four hits (44% hit rate) to NaV1.7, all with single digit micromolar binding affinities. Screening in both cases is completed in less than seven days. Finally, a high resolution X-ray crystallographic structure validates the predicted docking pose for the KLHDC2 ligand complex, demonstrating the effectiveness of our method in lead discovery.
- Open Access
Towards high-resolution modeling of small molecule-ion channel interactions
Ion channels are critical drug discovery targets for a wider range of pathologies, such as epilepsy, chronic pain, itch, autoimmunity, and cardiac arrhythmias. To develop effective and safe therapeutics, it is necessary to design small molecules with high potency and selectivity for specific ion channel subtypes. There has been increasing implementation of structure-guided drug design for the development of small molecules targeting ion channels. We evaluated the performance of two Rosetta ligand docking methods, RosettaLigand and GALigandDock, on structures of known ligand - cation channel complexes. Ligands were docked to voltage-gated sodium (NaV), voltage-gated calcium (CaV), and transient receptor potential vanilloid (TRPV) channel families. For each test case, RosettaLigand and GALigandDock methods were able to frequently sample a ligand binding pose within 1-2 Å root mean square deviation (RMSD) relative to the native ligand – channel structure. However, RosettaLigand and GALigandDock scoring functions cannot consistently identify native drug binding poses as top-scoring models. Our study reveals that the proper scoring criteria for RosettaLigand and GALigandDock modeling of ligand - ion channel complexes should be assessed on a case-by-case basis using sufficient ligand and receptor interface sampling, knowledge about state specific interactions of the ion channel and inherent receptor site flexibility that could influence ligand binding.
Open AccessTowards high-resolution modeling of small molecule-ion channel interactions
Ion channels are critical drug discovery targets for a wider range of pathologies, such as epilepsy, chronic pain, itch, autoimmunity, and cardiac arrhythmias. To develop effective and safe therapeutics, it is necessary to design small molecules with high potency and selectivity for specific ion channel subtypes. There has been increasing implementation of structure-guided drug design for the development of small molecules targeting ion channels. We evaluated the performance of two Rosetta ligand docking methods, RosettaLigand and GALigandDock, on structures of known ligand - cation channel complexes. Ligands were docked to voltage-gated sodium (NaV), voltage-gated calcium (CaV), and transient receptor potential vanilloid (TRPV) channel families. For each test case, RosettaLigand and GALigandDock methods were able to frequently sample a ligand binding pose within 1-2 Å root mean square deviation (RMSD) relative to the native ligand – channel structure. However, RosettaLigand and GALigandDock scoring functions cannot consistently identify native drug binding poses as top-scoring models. Our study reveals that the proper scoring criteria for RosettaLigand and GALigandDock modeling of ligand - ion channel complexes should be assessed on a case-by-case basis using sufficient ligand and receptor interface sampling, knowledge about state specific interactions of the ion channel and inherent receptor site flexibility that could influence ligand binding.
- Open Access
An artificial intelligence accelerated virtual screening platform for drug discovery
Structure-based virtual screening is a key tool in early drug discovery, with growing interest in the screening of multi-billion chemical compound libraries. However, the success of virtual screening crucially depends on the accuracy of the binding pose and binding affinity predicted by computational docking. Here we developed a highly accurate structure-based virtual screen method, RosettaVS, for predicting docking poses and binding affinities. Our approach outperforms other state-of-the-art methods on a wide range of benchmarks, partially due to our ability to model receptor flexibility. We incorporate this into a new open-source artificial intelligence accelerated virtual screening platform for drug discovery. Using this platform, we screened multi-billion compound libraries against two unrelated targets, a novel ubiquitin ligase target KLHDC2 and the human voltage-gated sodium channel Na V 1.7. On both targets, we discover hits, including seven novel hits (14% hit rate) to KLHDC2 and four novel hits (44% hit rate) to Na V 1.7 with single digit micromolar binding affinities. Screening in both cases was completed in less than seven days. Finally, a high resolution X-ray crystallographic structure validates the predicted docking pose for the KLHDC2 ligand complex, demonstrating the effectiveness of our method in lead discovery.
Open AccessAn artificial intelligence accelerated virtual screening platform for drug discovery
Structure-based virtual screening is a key tool in early drug discovery, with growing interest in the screening of multi-billion chemical compound libraries. However, the success of virtual screening crucially depends on the accuracy of the binding pose and binding affinity predicted by computational docking. Here we developed a highly accurate structure-based virtual screen method, RosettaVS, for predicting docking poses and binding affinities. Our approach outperforms other state-of-the-art methods on a wide range of benchmarks, partially due to our ability to model receptor flexibility. We incorporate this into a new open-source artificial intelligence accelerated virtual screening platform for drug discovery. Using this platform, we screened multi-billion compound libraries against two unrelated targets, a novel ubiquitin ligase target KLHDC2 and the human voltage-gated sodium channel Na V 1.7. On both targets, we discover hits, including seven novel hits (14% hit rate) to KLHDC2 and four novel hits (44% hit rate) to Na V 1.7 with single digit micromolar binding affinities. Screening in both cases was completed in less than seven days. Finally, a high resolution X-ray crystallographic structure validates the predicted docking pose for the KLHDC2 ligand complex, demonstrating the effectiveness of our method in lead discovery.
- Open Access
- 1 March 2023
- Published by Springer Nature in Nature Communications
- Vol. 14 (1) , 1-10
Automatic and accurate ligand structure determination guided by cryo-electron microscopy maps
Advances in cryo-electron microscopy (cryoEM) and deep-learning guided protein structure prediction have expedited structural studies of protein complexes. However, methods for accurately determining ligand conformations are lacking. In this manuscript, we develop EMERALD, a tool for automatically determining ligand structures guided by medium-resolution cryoEM density. We show this method is robust at predicting ligands along with surrounding side chains in maps as low as 4.5 Å local resolution. Combining this with a measure of placement confidence and running on all protein/ligand structures in the EMDB, we show that 57% of ligands replicate the deposited model, 16% confidently find alternate conformations, 22% have ambiguous density where multiple conformations might be present, and 5% are incorrectly placed. For five cases where our approach finds an alternate conformation with high confidence, high-resolution crystal structures validate our placement. EMERALD and the resulting analysis should prove critical in using cryoEM to solve protein-ligand complexes.
- 1 March 2023
- Published by Springer Nature in Nature Communications
- Vol. 14 (1) , 1-10
Open AccessAutomatic and accurate ligand structure determination guided by cryo-electron microscopy maps
Advances in cryo-electron microscopy (cryoEM) and deep-learning guided protein structure prediction have expedited structural studies of protein complexes. However, methods for accurately determining ligand conformations are lacking. In this manuscript, we develop EMERALD, a tool for automatically determining ligand structures guided by medium-resolution cryoEM density. We show this method is robust at predicting ligands along with surrounding side chains in maps as low as 4.5 Å local resolution. Combining this with a measure of placement confidence and running on all protein/ligand structures in the EMDB, we show that 57% of ligands replicate the deposited model, 16% confidently find alternate conformations, 22% have ambiguous density where multiple conformations might be present, and 5% are incorrectly placed. For five cases where our approach finds an alternate conformation with high confidence, high-resolution crystal structures validate our placement. EMERALD and the resulting analysis should prove critical in using cryoEM to solve protein-ligand complexes.
- Open Access
Cyclic peptide structure prediction and design using AlphaFold
Deep learning networks offer considerable opportunities for accurate structure prediction and design of biomolecules. While cyclic peptides have gained significant traction as a therapeutic modality, developing deep learning methods for designing such peptides has been slow, mostly due to the small number of available structures for molecules in this size range. Here, we report approaches to modify the AlphaFold network for accurate structure prediction and design of cyclic peptides. Our results show this approach can accurately predict the structures of native cyclic peptides from a single sequence, with 36 out of 49 cases predicted with high confidence (pLDDT > 0.85) matching the native structure with root mean squared deviation (RMSD) less than 1.5 Å. Further extending our approach, we describe computational methods for designing sequences of peptide backbones generated by other backbone sampling methods and for de novo design of new macrocyclic peptides. We extensively sampled the structural diversity of cyclic peptides between 7–13 amino acids, and identified around 10,000 unique design candidates predicted to fold into the designed structures with high confidence. X-ray crystal structures for seven sequences with diverse sizes and structures designed by our approach match very closely with the design models (root mean squared deviation < 1.0 Å), highlighting the atomic level accuracy in our approach. The computational methods and scaffolds developed here provide the basis for custom-designing peptides for targeted therapeutic applications.
Open AccessCyclic peptide structure prediction and design using AlphaFold
Deep learning networks offer considerable opportunities for accurate structure prediction and design of biomolecules. While cyclic peptides have gained significant traction as a therapeutic modality, developing deep learning methods for designing such peptides has been slow, mostly due to the small number of available structures for molecules in this size range. Here, we report approaches to modify the AlphaFold network for accurate structure prediction and design of cyclic peptides. Our results show this approach can accurately predict the structures of native cyclic peptides from a single sequence, with 36 out of 49 cases predicted with high confidence (pLDDT > 0.85) matching the native structure with root mean squared deviation (RMSD) less than 1.5 Å. Further extending our approach, we describe computational methods for designing sequences of peptide backbones generated by other backbone sampling methods and for de novo design of new macrocyclic peptides. We extensively sampled the structural diversity of cyclic peptides between 7–13 amino acids, and identified around 10,000 unique design candidates predicted to fold into the designed structures with high confidence. X-ray crystal structures for seven sequences with diverse sizes and structures designed by our approach match very closely with the design models (root mean squared deviation < 1.0 Å), highlighting the atomic level accuracy in our approach. The computational methods and scaffolds developed here provide the basis for custom-designing peptides for targeted therapeutic applications.
- Open Access
- 12 February 2021
- Published by American Chemical Society (ACS) in Journal of Chemical Theory and Computation
- Vol. 17 (3) , 2000-2010
Force Field Optimization Guided by Small Molecule Crystal Lattice Data Enables Consistent Sub-Angstrom Protein–Ligand Docking
Accurate and rapid calculation of protein-small molecule interaction free energies is critical for computational drug discovery. Because of the large chemical space spanned by drug-like molecules, classical force fields contain thousands of parameters describing atom-pair distance and torsional preferences; each parameter is typically optimized independently on simple representative molecules. Here, we describe a new approach in which small molecule force field parameters are jointly optimized guided by the rich source of information contained within thousands of available small molecule crystal structures. We optimize parameters by requiring that the experimentally determined molecular lattice arrangements have lower energy than all alternative lattice arrangements. Thousands of independent crystal lattice-prediction simulations were run on each of 1386 small molecule crystal structures, and energy function parameters of an implicit solvent energy model were optimized, so native crystal lattice arrangements had the lowest energy. The resulting energy model was implemented in Rosetta, together with a rapid genetic algorithm docking method employing grid-based scoring and receptor flexibility. The success rate of bound structure recapitulation in cross-docking on 1112 complexes was improved by more than 10% over previously published methods, with solutions within <1 Å in over half of the cases. Our results demonstrate that small molecule crystal structures are a rich source of information for guiding molecular force field development, and the improved Rosetta energy function should increase accuracy in a wide range of small molecule structure prediction and design studies.
- 12 February 2021
- Published by American Chemical Society (ACS) in Journal of Chemical Theory and Computation
- Vol. 17 (3) , 2000-2010
Open AccessForce Field Optimization Guided by Small Molecule Crystal Lattice Data Enables Consistent Sub-Angstrom Protein–Ligand Docking
Accurate and rapid calculation of protein-small molecule interaction free energies is critical for computational drug discovery. Because of the large chemical space spanned by drug-like molecules, classical force fields contain thousands of parameters describing atom-pair distance and torsional preferences; each parameter is typically optimized independently on simple representative molecules. Here, we describe a new approach in which small molecule force field parameters are jointly optimized guided by the rich source of information contained within thousands of available small molecule crystal structures. We optimize parameters by requiring that the experimentally determined molecular lattice arrangements have lower energy than all alternative lattice arrangements. Thousands of independent crystal lattice-prediction simulations were run on each of 1386 small molecule crystal structures, and energy function parameters of an implicit solvent energy model were optimized, so native crystal lattice arrangements had the lowest energy. The resulting energy model was implemented in Rosetta, together with a rapid genetic algorithm docking method employing grid-based scoring and receptor flexibility. The success rate of bound structure recapitulation in cross-docking on 1112 complexes was improved by more than 10% over previously published methods, with solutions within <1 Å in over half of the cases. Our results demonstrate that small molecule crystal structures are a rich source of information for guiding molecular force field development, and the improved Rosetta energy function should increase accuracy in a wide range of small molecule structure prediction and design studies.