CoChat

The maximal and current accuracy of rigorous protein-ligand binding free energy calculations

Open access

Published version of a preprint

The DOI matches its Crossref record; 90 of the 90 references it cites were found in a registry or an index.

Abstract

Computational techniques can speed up the identification of hits and accelerate the development of candidate molecules for drug discovery. Among techniques for predicting relative binding affinities, the most consistently accurate is free energy perturbation (FEP), a class of rigorous physics-based methods. However, uncertainty remains about how accurate FEP is and can ever be. Here, we present what we believe to be the largest publicly available dataset of proteins and congeneric series of small molecules, and assess the accuracy of the leading FEP workflow. To ascertain the limit of achievable accuracy, we also survey the reproducibility of experimental relative affinity measurements. We find a wide variability in experimental accuracy and a correspondence between binding and functional assays. When careful preparation of protein and ligand structures is undertaken, FEP can achieve accuracy comparable to experimental reproducibility. Throughout, we highlight reliable protocols that can help maximize the accuracy of FEP in prospective studies.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s42004-023-01019-9

Cites

90 of 90 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Examples from the FEP benchmark where the calculations were augmented. Where possible, FEP was used to automatically determine the (a) preferred rotamer state of R-group modifications (like with MCL1), b protonation state of the ligand (like with TNKS2), and (c) the protonation state of side chain residues (like with PTP1B). d In some cases, (like CK2), atom mappings were altered to enhanced conformation sampling. MCL1 exemplifies cases where multiple orientations of R-groups were added to the FEP map and post-processed based on Supplementary Eq. 2 in the SI. TNKS2 exemplifies cases where multiple protonation states were post processed based on our pKa correction workflow55. In PTP1B, the sulfur atom is in close proximity to a number of backbone NH groups (not shown) which are likely to be the the main drivers of the predicted negatively charged protonation state.
Fig. 1 Examples from the FEP benchmark where the calculations were augmented. Where possible, FEP was used to automatically determine the (a) preferred rotamer state of R-group modifications (like with MCL1), b protonation state of the ligand (like with TNKS2), and (c) the protonation state of side chain residues (like with PTP1B). d In some cases, (like CK2), atom mappings were altered to enhanced conformation sampling. MCL1 exemplifies cases where multiple orientations of R-groups were added to the FEP map and post-processed based on Supplementary Eq. 2 in the SI. TNKS2 exemplifies cases where multiple protonation states were post processed based on our pKa correction workflow55. In PTP1B, the sulfur atom is in close proximity to a number of backbone NH groups (not shown) which are likely to be the the main drivers of the predicted negatively charged protonation state.CC BY · PubMed Central
Fig. 2 Scatter plots showing the range of agreement of ΔGs between different experimental assays and FEP+ predictions. Panels a–c compare measured affinities between two different experiments and panels d–f compare FEP+ predictions (y-axis) against experimental measurements (x-axis). Panels a and d show examples where the pairwise RMSE of relative binding free energies was much better than average, panels b and e show examples where the RMSE was close to the average, and panels c and f show examples where the RMSE was worse than average. The top left of each plot shows Kendall τ and pairwise RMSE for each data set. Points in the dark gray area are measurements or predictions that are within 1 kcal mol−1 of each other, and points in the light gray area agree within 2 kcal mol−1. a Shows that isothermal titration calorimetry (ITC) and fluorescence polarization (FP) binding free energy measurements of galectin ligands are offset by ~1 kcal mol−1 - this offset is irrelevant for rank ordering and does not affect the correlation or the pairwise RMSE metric. The offset of the FEP+ ΔG predictions was determined by ensuring the mean of the ΔGs was equal to the mean of experimental ΔGs on the x-axis. Where the data was available, we included the reported standard error of the experimental measurements; the standard errors from FEP+ as calculated with the cycle-closure algorithm are also indicated on the bottom row of plots.
Fig. 2 Scatter plots showing the range of agreement of ΔGs between different experimental assays and FEP+ predictions. Panels a–c compare measured affinities between two different experiments and panels d–f compare FEP+ predictions (y-axis) against experimental measurements (x-axis). Panels a and d show examples where the pairwise RMSE of relative binding free energies was much better than average, panels b and e show examples where the RMSE was close to the average, and panels c and f show examples where the RMSE was worse than average. The top left of each plot shows Kendall τ and pairwise RMSE for each data set. Points in the dark gray area are measurements or predictions that are within 1 kcal mol−1 of each other, and points in the light gray area agree within 2 kcal mol−1. a Shows that isothermal titration calorimetry (ITC) and fluorescence polarization (FP) binding free energy measurements of galectin ligands are offset by ~1 kcal mol−1 - this offset is irrelevant for rank ordering and does not affect the correlation or the pairwise RMSE metric. The offset of the FEP+ ΔG predictions was determined by ensuring the mean of the ΔGs was equal to the mean of experimental ΔGs on the x-axis. Where the data was available, we included the reported standard error of the experimental measurements; the standard errors from FEP+ as calculated with the cycle-closure algorithm are also indicated on the bottom row of plots.CC BY · PubMed Central
Fig. 3 Summarizing the error distributions in the experimental survey and FEP benchmark. a Boxplots comparing the root-mean-square error (RMSE) between relative binding free energies from different experimental assays (purple) and the FEP+ predictions against experimental data (green). The top and bottom of the boxes represent the 25th and 75th percentiles and the dark line represents the median. The whiskers extend to a maximum of 1.5 times the interquartile range. Circles are the RMSEs from comparing two experimental assays (green) or the RMSEs of a FEP+ perturbation graph (purple). The size of each circle is proportional to the number of ligands in the series in either an assay comparison or perturbation graph. The two largest data points in the experimental survey are from the COVID moonshot project68 and project A from Supplementary Table 4. The median RMSE in the experimental survey is 0.85 kcal mol−1 and the median in the FEP+ benchmark is 1.08 kcal mol−1. b All pairwise relative binding free energy differences from the experimental survey and all pairwise FEP+ errors. The histograms were symmetrized about the x=0 line in the sense that all N × (N − 1) pairs of compounds were used. The error distributions are bell-shaped and can be approximated by t-distributions.
Fig. 3 Summarizing the error distributions in the experimental survey and FEP benchmark. a Boxplots comparing the root-mean-square error (RMSE) between relative binding free energies from different experimental assays (purple) and the FEP+ predictions against experimental data (green). The top and bottom of the boxes represent the 25th and 75th percentiles and the dark line represents the median. The whiskers extend to a maximum of 1.5 times the interquartile range. Circles are the RMSEs from comparing two experimental assays (green) or the RMSEs of a FEP+ perturbation graph (purple). The size of each circle is proportional to the number of ligands in the series in either an assay comparison or perturbation graph. The two largest data points in the experimental survey are from the COVID moonshot project68 and project A from Supplementary Table 4. The median RMSE in the experimental survey is 0.85 kcal mol−1 and the median in the FEP+ benchmark is 1.08 kcal mol−1. b All pairwise relative binding free energy differences from the experimental survey and all pairwise FEP+ errors. The histograms were symmetrized about the x=0 line in the sense that all N × (N − 1) pairs of compounds were used. The error distributions are bell-shaped and can be approximated by t-distributions.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 90 references

Research works printed but not linked: 1

  1. Peterson, K. Molecular basis for galectin-ligand interactions. Design, Synthesis And Analysis. Ph.D. thesis (Lund University, 2018).

Cited by

See all 211

Metadata

Authors
Ross, G. A., Lu, C., Scarabelli, G., Albanese, S. K., Houang, E., Abel, R., Harder, E. D., & Wang, L.
Affiliations
Schrodinger (United States) (US)
Corresponding author
  • Ross, Gregory A.
  • Wang, Lingle
Venue
Communications Chemistry
Published
2023-10-14
Volume
6
Issue
1
Pages
222-222
ISSN
2399-3669 (ISSN-L)
OpenAlex
W4387639561
Cited by
178
References
87
Citations by year
2026: 58 · 2025: 79 · 2024: 35 · 2023: 5
FWCI
25.48
Type
journal article
Language
en
Open access
Licence
Locations
  • Communications Chemistry (publishedVersion, cc-by)
  • Research Square (acceptedVersion, cc-by)
  • ChemRxiv (acceptedVersion, cc-by-nc)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Life Sciences › Pharmacology, Toxicology and Pharmaceutics › Drug Discovery › Computational Drug Discovery Methods
  • Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Advanced Chemical Physics Studies
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Protein Structure and Dynamics
Keywords
free energy perturbation, protein-ligand binding affinity, congeneric series, experimental reproducibility, drug discovery, accuracy assessment

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: The maximal and current accuracy of rigorous protein-ligand binding free energy calculations. https://app.cochat.ai/papers/10.1038/s42004-023-01019-9
The paper
Ross, G. A., Lu, C., Scarabelli, G., Albanese, S. K., Houang, E., Abel, R., Harder, E. D., & Wang, L. (2023). The maximal and current accuracy of rigorous protein-ligand binding free energy calculations. Communications Chemistry. https://doi.org/10.1038/s42004-023-01019-9

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.