Pith. sign in

REVIEW 3 major objections 4 minor 52 references

Automated and optimally FRET-assisted structural modeling

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Optimally chosen FRET distance measurements, as few as 13 to 23 per protein, can drive multiscale models to near-crystallographic accuracy for dynamic multi-domain proteins, this study argues.

desk verdict A substantial FRET-assisted modeling workflow with a real experimental test, but the absolute quality metric rests on a self-referential complexity estimate that needs independent validation. read the letter →

arxiv 1909.02148 v1 pith:3CZNZPCQ submitted 2019-09-04 q-bio.QM physics.bio-phq-bio.BM

Mykola Dimura , Thomas O. Peulen , Hugo Sanabria , Dmitro Rodnin , Katherina Hemmen , Claus A.M. Seidel , Holger Gohlke This is my paper · ORCID
classification q-bio.QMphysics.bio-phq-bio.BM
keywords FRETproteinstructuredeterminationintegrativestructuralbiologyoptimalpairselectionnormalizedchi-squaredqualitymetricmultiscalemodelingFRET-restrainedmoleculardynamicsaccessiblevolume
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper aims to show that a small set of optimally chosen FRET distance measurements, combined with multiscale computational sampling, can determine protein structures to near-crystallographic accuracy in a largely automated pipeline. The authors introduce a pair-selection algorithm that chooses donor-acceptor labeling sites to maximize expected structural precision, and a normalized goodness-of-fit statistic, $\chi_n^2$, that they argue plays the role for FRET models that the free R-factor plays in crystallography. On a benchmark of six proteins, using 13 to 23 FRET measurements, they report selected ensembles with precision of 2 to 3.5 Å and accuracy of 1.8 to 3.5 Å against X-ray target structures. If correct, this would make FRET a practical route to state-specific structures of dynamic multi-domain proteins that resist crystallization, using only a handful of measurements.

What carries the argument

The argument rides on three linked objects. First is the expected-precision estimator $\langle\langle \mathrm{RMSD}\rangle\rangle$, a double average over the prior ensemble: for each reference conformer it simulates the FRET observables it would produce, p-values for all other conformers are computed using the chi-squared test, and the RMSD matrix is weighted by those p-values; the pair-selection algorithm greedily chooses the donor-acceptor pairs that minimize this quantity. Second is the normalized chi-squared $\chi_n^2$, defined as $\chi^2 / \mathrm{Inv.}\chi^2(p=0.68, N_{\mathrm{dof}})$, which puts every model on the same statistical scale regardless of the number of measurements. Third is the automatic estimate of model complexity $N_{\mathrm{SMP}}$: starting from zero, the pair-selection algorithm's own chosen number of FRET pairs is taken as the number of independent coordinates, so that $N_{\mathrm{dof}} = N_{\mathrm{measurements}} - N_{\mathrm{SMP}}$; this makes $\chi_n^2$ usable as an $R_{\mathrm{free}}$-like cross-validation score. These are wrapped in sampling engines: FRET-guided normal-mode geometric sampling and FRET-restrained molecular dynamics in which pseudo-atoms placed at the mean accessible-volume position of each dye carry the experimental restraints.

What would settle it

Run the workflow on a rigid-body ensemble whose true number of independent coordinates is known analytically (six degrees of freedom) and check that the pair-selection estimator returns six; then, in a blinded benchmark with simulated FRET data from a known crystal structure, check that conformers with $\chi_n^2<1$ are predominantly below 2 Å RMSD from the target rather than above 5 Å. A failure of either check would remove the grounding for the absolute quality scale.

Watch

Extended reading notes

Core claim

On its own terms, the paper claims that accurate, efficient, and largely automated protein structure determination is possible from optimally designed FRET experiments combined with multiscale structural modeling. The discovery is packaged as a six-step iterative workflow: collect prior knowledge, generate an initial conformational ensemble, select the most informative FRET pairs, measure them, screen the ensemble with an absolute quality criterion, and refine by FRET-guided sampling. The central quantitative claim is that the normalized statistic $\chi_n^2 = \chi^2 / \mathrm{Inv.}\chi^2(p=0.68, N_{\mathrm{dof}})$ provides an absolute, cross-validation-type measure of model quality analogous to $R_{\mathrm{free}}$, provided the number of degrees of freedom $N_{\mathrm{dof}} = N_{\mathrm{measurements}} - N_{\mathrm{SMP}}$ is properly estimated. For six benchmark proteins, FRET-consistent ensembles selected by $\chi_n^2 < 1$ have precision 2 to 3.5 Å and accuracy 1.8 to 3.5 Å against the target X-ray structure, using as few as 13 to 23 FRET measurements depending on the prior ensemble. This includes an experimental case, T4 lysozyme, where two conformers only 4 Å apart are distinguished.

Load-bearing premise

The absolute quality scale rests on a self-consistent guess: the model's number of independent coordinates is set equal to the number of FRET pairs the selection algorithm chooses, and if that guess undercounts or overcounts the true complexity, every $\chi_n^2$ value and p-value—and with them the claimed precision and accuracy numbers—would no longer be on an absolute scale.

Editorial extensions

If this is right

  • A handful of measurements—13 to 23 FRET pairs—can replace extensive sampling or brute-force experimental coverage for multi-domain proteins, if the prior ensemble is diverse enough.
  • Because $\chi_n^2$ is an absolute quality measure, FRET-selected ensembles can be compared across different proteins and different numbers of measurements without rescaling, analogous to how $R_{\mathrm{free}}$ is used in crystallography.
  • The limiting $\chi_n^2$ value reached by different seed structures exposes errors in the starting fold, so the workflow doubles as a diagnostic of which prior model is wrong and how wrong.
  • The implicit-dye pseudo-atom restraint scheme avoids explicit dye simulation and its convergence problems, making FRET-restrained molecular dynamics practical for routine refinement.
  • The pair-selection and quality-scoring machinery transfers with minimal changes to other label-based methods such as EPR, paramagnetic relaxation enhancement NMR, and vibrational spectroscopy.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the self-consistent $N_{\mathrm{SMP}}$ estimate is sound, the same heuristic could supply a model-complexity score for any ensemble-based integrative modeling pipeline, not just FRET, since it needs only a conformational ensemble and a distance-like observable.
  • The reported accuracy range suggests that combining this workflow with modern deep-learning structure predictors could cut the required number of measurements even further; a natural test is to run the pair-selection algorithm on predictor-generated ensembles and compare the pair counts needed for the same precision.
  • Because the quality score depends on accurate experimental error estimates, remaining accuracy gaps may reflect dye linker and orientation-factor errors rather than sampling; this could be tested by adding explicit dye simulations where FRET accuracy plateaus above 2 Å.
  • If the workflow scales to larger complexes, it would offer a route to state-specific structures of transient assemblies in solution, where crystallization and cryo-electron microscopy averaging are difficult; the T4 lysozyme case with two 4-microsecond-lived conformers is a first indication.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. This paper presents an integrative workflow for automated FRET-assisted protein structure modeling. A prior structural ensemble is generated by normal-mode-based geometric simulations (NMSim) from homology/CASP seeds; a feature-selection algorithm proposes an optimal set of FRET donor-acceptor pairs by minimizing the expected precision ⟨⟨RMSD⟩⟩; experimental FRET distances are then used to screen the ensemble with a normalized chi-squared statistic χ_n^2 and to drive further sampling via guided NMSim and FRET-restrained MD with pseudo-atom representations of dye accessible volumes. The method is benchmarked on six proteins: five with simulated FRET data and T4 lysozyme with experimental single-molecule FRET data. The authors report FRET-selected ensembles with precision 2–3.5 Å and accuracy 1.8–3.5 Å against X-ray targets using 13–23 FRET pairs, and they propose the χ_n^2 statistic as an absolute, Rfree-like quality measure.

Significance. If the accuracy claims hold, this is an important advance: a small number of information-theoretically chosen FRET restraints could drive multi-scale models to near-crystallographic accuracy, and the automated pair design would be practical for dynamic multi-domain systems. The paper has several genuine strengths: it makes software available (Olga, NMSim, FRETrest/LabelLib), it uses a real experimental T4L dataset as an independent check, and it is unusually explicit about the heuristic nature of its complexity estimate. However, the absolute quality interpretation of χ_n^2 depends on a self-consistently estimated number of degrees of freedom, N_SMP, and the simulated benchmarks share the same accessible-volume forward model between data generation and screening. These issues are central because they directly affect the reported RMSD ranges in Table 4; they are addressable by additional validation experiments and sensitivity analyses.

major comments (3)
  1. [Online Methods §5 (Eqs. 3, 8)] Online Methods §5 (Eq. 3): the estimate N_SMP = number of FRET pairs selected by the algorithm is self-referential. Because the pair-selection p-values and χ_n^2 values depend on N_dof = N_measurements − N_SMP (Eq. 3), the “model complexity” is not an independent property of the structural ensemble, and the claim that Eq. (8) is an absolute quality measure is not established. Table 4 reports RMSDs of structures passing the χ_n^2 < 1 filter, so the headline 1.8–3.5 Å numbers inherit this calibration. The authors should validate N_SMP against an independent complexity estimator (e.g., principal-component or fluctuation-spectrum dimension of the ensemble, or the known analytical N_SMP for rigid-body models) and report how Table 4 changes when N_SMP is varied by a factor of two.
  2. [Online Methods §3 and Supplementary Note 5] Simulated benchmarks use the same accessible-volume (AV) forward model for generating reference distances and for screening/restraining structures. This shared model can only overestimate agreement, because the inverse screening and pair selection use the same AV pseudo-atom representation whose systematic errors (linker parameterization, AV mean-position approximation) cancel by construction. The T4L experiments are the only independent check. The authors should add a sensitivity test in which synthetic FRET data are generated with an independent forward model (e.g., explicit-dye MD or a perturbed AV parameterization) and show that the 2–3 Å accuracy claims survive.
  3. [Online Methods Eq. 8 and Fig. 1i] The claim that χ_n^2 is analogous to R_free and detects wrong seed folds (Fig. 1i) needs a direct falsifiability test on the absolute scale. For fixed N_dof, χ_n^2 can rank conformers, but the absolute threshold χ_n^2 < 1 is calibrated through N_SMP; if N_SMP is biased low, decoys pass, and if biased high, valid conformers are rejected. The paper should report a decoy test, for example the distribution of χ_n^2 for ensembles known to be wrong (e.g., the discarded CASP models or other PDB conformers of the benchmark proteins), and show that the χ_n^2 < 1 selection has a low false-positive rate.
minor comments (4)
  1. [Online Methods Eqs. 8–9] Eq. (9) is called the inverse chi-squared distribution but the displayed expression is a probability density function; a quantile function is needed for the normalization in Eq. (8). Please define Inv.chi^2(p,N_dof) unambiguously and provide a reference or derivation.
  2. [Online Methods §1] The word “sheer” should be “shear” in the description of internal motions.
  3. [Code availability] Providing a shared guest account in an arXiv posting is not a durable archival mechanism; the repositories should be released under version tags and deposited in a permanent archive before publication.
  4. [Table 4] Please state explicitly, for each row, the total number N_measurements and the resulting N_dof used in Eq. (3); the split into guiding and validation pairs is crucial because for FRET-restrained models the validation pairs are reportedly the only contributors to N_dof.

Circularity Check

1 steps flagged · score 6.0 of 10

The absolute χ_n² quality metric is calibrated by a self-referential N_SMP estimate: N_SMP is defined as the number of pairs selected by the same algorithm whose significance tests depend on N_dof = N_measurements − N_SMP.

  1. self definitional [Online Methods, section 5 and section 2 (Eqs. 3, 4, 6, 8); Table 4]
    "Initially, to obtain an N_SMP estimate, we start by assuming N_SMP,0 = 0, and determine a set of DA pairs needed to describe the conformations within an ensemble with a desired precision ⟨⟨RMSD⟩⟩ employing our DA pair selection algorithm. Each DA pair can be seen as a coordinate, and the number of DA pairs corresponds to our definition of N_SMP."

    N_SMP is not counted from the mechanical degrees of freedom of the model; it is defined as the number of FRET pairs that the pair-selection algorithm selects. That algorithm's expected-precision score ⟨⟨RMSD⟩⟩ (Eq. 6) is a p-value-weighted average, and the p-values (Eq. 4) depend on N_dof = N_measurements − N_SMP (Eq. 3). Thus the quantity that sets the degrees of freedom is itself an output of the same statistic it calibrates. The paper even starts the estimate with N_SMP,0 = 0, i.e., it temporarily pretends every FRET observable is an independent coordinate; re-running with the resulting count does not break the self-reference. For FRET-restrained models, N_SMP is then equated to the number of FRET restraints, leaving only 3–10 validation pairs as N_dof in Table 4.

full rationale

The paper has substantial external content: Table 4 reports RMSD values of FRET-selected structures against known X-ray targets, and the T4L results use real experimental FRET data, so the accuracy figures are not manufactured. The circularity is confined to the internal calibration of the absolute quality measure. The derivation chain for χ_n² relies on Eq. 3, N_dof = N_measurements − N_SMP, and Online Methods section 5 defines N_SMP as the number of DA pairs selected by the same algorithm whose p-values and expected precision depend on N_dof. For FRET-restrained models the identification N_SMP = N_FRET_restraints makes the screening degrees of freedom equal to only the held-out validation pairs. This does not destroy the relative ordering of conformers for fixed N_dof, and the benchmark accuracies may still be correct, but the claim that χ_n² is an absolute, cross-validated quality measure is not derived from first principles; it is calibrated by the heuristic it is supposed to validate. The paper explicitly concedes the heuristic nature, and the only supporting checks are internal (Fig. 1i) or depend on known benchmark targets (Supplementary Fig. 6), which are unavailable in genuine prediction use. Score 6 reflects one central, partially circular calibration step while acknowledging the independent external benchmark.

Assumptions & free parameters 6 free parameters · 6 assumptions · 1 invented entities

The workflow rests on statistical assumptions about FRET error distributions, a specific forward model for dye distances, and a heuristic for model complexity that is estimated from the same algorithm it feeds. The benchmark results support the workflow's usefulness but do not independently validate the N_SMP estimate or the forward model. Most free parameters are standard simulation settings or empirically tuned restraint parameters.

free parameters (6)
  • Fmax force cap = 50 pN
    Empirically determined cap on FRET-restraint force in MD (Online Methods section 8); central to restraint implementation but not derived.
  • Annealing schedule kT range = 0 to 1 chi_n^2 units
    Chosen schedule for Metropolis acceptance in FRET-guided NMSim (Supplementary Note 6), not derived from theory.
  • Number of NMSim normal modes = 10
    Linear combination of first 10 modes used to bias backbone motions (Online Methods section 6); a default from the NMSim method rather than fitted here.
  • N_SMP model complexity estimate = Variable, set by pair-selection algorithm
    Estimated via the heuristic in Online Methods section 5: the number of FRET pairs the selection algorithm requires starting from N_SMP=0; this feeds into N_dof and chi_n^2, so the quality metric depends on it.
  • Convergence RMSD threshold = 3 Å
    Chosen criterion RMSDij < 3 Å for workflow convergence (Main text step 5).
  • YaaA seed selection by hand = 10 of 100 cluster representatives
    Hand-selected from 100 clusters (Online Methods section 1); a subjective choice in the benchmark, not an algorithmically fixed set.
assumptions (6)
  • domain assumption Chi-squared and inverse chi-squared distribution formulas (Eqs. 4, 5, 9) are the correct statistical model for FRET distance errors.
    Used throughout the quality metric; assumes Gaussian-distributed distance errors with known per-pair uncertainties.
  • domain assumption The Accessible Volume (AV) model of Kalinin et al. 2012 accurately converts protein structure into FRET observables.
    Invoked in Online Methods section 3 for screening and for generating simulated reference data; the same model is used both forward and inverse.
  • ad hoc to paper Pseudo-atoms placed at the mean accessible-volume position can represent FRET distances accurately despite being anchored to backbone and not explicitly diffusing.
    Online Methods section 8: this is the core approximation of the FRET-restrained MD protocol, justified only by the benchmark results.
  • domain assumption NMSim normal-mode sampling with default parameters adequately covers the relevant conformational space when unrestrained.
    Online Methods section 6: the entire prior ensemble generation depends on this coverage claim.
  • ad hoc to paper The heuristic N_SMP estimation converges to the true number of independent coordinates of a structural model.
    Online Methods section 5: the feedback between pair selection and N_dof is asserted to work, supported only by internal consistency (Fig. 1i).
  • domain assumption AMBER ff14SB force field and TIP3P water give accurate MD sampling for restrained refinement.
    Online Methods section 8: standard simulation assumptions for atomistic refinement.
invented entities (1)
  • FRET pseudo-atoms
    purpose: Anchor implicit fluorophore positions to the backbone and carry harmonic-linear FRET restraints in MD simulations, avoiding explicit dye simulation.
    Pseudo-atoms are a computational construct with no standalone falsifiable prediction; the benchmark accuracies provide indirect validation but no independent evidence for the entity itself.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Automated and optimally FRET-assisted structural modeling." pith.science (2026). https://pith.science/paper/3CZNZPCQ

@misc{pith2026190902148,
  author       = {Pith},
  title        = {Pith review of: Automated and optimally FRET-assisted structural modeling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3CZNZPCQ}},
  note         = {Machine review of arXiv:1909.02148}
}
read the original abstract

FRET experiments can yield state-specific structural information on complex dynamic biomolecular assemblies. However, FRET experiments need to be combined with computer simulations to overcome their sparsity. We introduce (i) an automated FRET experiment design tool determining optimal FRET pairs for structural modeling, (ii) a protocol for efficient FRET-assisted computational structural modeling at multiple scales, and (iii) a quantitative quality estimate for judging the accuracy of determined structures. We tested against simulated and experimental data.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 52 canonical work pages

  1. [1]

    E., Chivian, D

    Kim, D. E., Chivian, D. & Baker, D. Protein structure predic tion and analysis using the Robetta server. Nucleic Acids Res. 32, W526‐W531 (2004)

    work page 2004

  2. [2]

    Kelley, L. A. & Sternberg, M. J. E. Protein structure predic tion on the Web: a case study using the Phyre server. Nature Protocols 4, 363‐371 (2009)

    work page 2009

  3. [3]

    & Sali, A

    Webb, B. & Sali, A. Protein structure modeling with MODELLER . Methods Mol. Biol. 1137, 1‐15 (2014)

    work page 2014

  4. [4]

    I‐TASSER server for protein 3D structure predictio n

    Zhang, Y. I‐TASSER server for protein 3D structure predictio n. BMC Bioinformatics 9 (2008)

    work page 2008

  5. [5]

    & Schwede, T

    Arnold, K., Bordoli, L., Kopp, J. & Schwede, T. The SWISS‐MO D E L w o r k s p a c e : a w e b ‐ b a s e d e n v i r o n m e n t f o r p r o t e i n structure homology modelling. Bioinformatics 22, 195‐201 (2006)

    work page 2006

  6. [6]

    & Kern, D

    Henzler‐Wildman, K. & Kern, D. Dynamic personalities of proteins. Nature 450, 964‐972 (2007)

    work page 2007

  7. [7]

    & Kay, L

    Neudecker, P., Robustelli, P., Cavalli, A., Walsh, P., Lunds trom, P., Zarrine‐Afsar, A., Sharpe, S., Vendruscolo, M. & Kay, L. E. Structure of an intermediate state in protein folding and aggregation. Science 336, 362‐366 (2012). 8 . D i m u r a , M . , P e u l e n , T . O . , H a n k e , C . A . , P r a k a s h , A . , G o h l k e, H . & S e i d e l , C...

    work page 2012

  8. [9]

    & H ugel, T

    Hellenkamp, B., Wortmann, P., Kandzia, F., Zacharias, M. & H ugel, T. Multidomain Structure and Correlated Dynamics Determined by Self‐Consistent FRET Networks. Nat. Meth. 14, 174‐180 (2017)

    work page 2017

Show all 52 references
  1. [10]

    F., Felekyan, S., Haustein, E., König, M., Fasshauer, D., Grubmüller, H., Jahn, R

    Margittai, M., Widengren, J., Schweinberger, E., Schröder, G. F., Felekyan, S., Haustein, E., König, M., Fasshauer, D., Grubmüller, H., Jahn, R. & Seidel, C. A. M. Single‐molecule flu orescence resonance energy transfer reveals a dynamic equilibrium between closed and open con...

    2003

  2. [11]

    & Michaelis, J

    Muschielok, A., Andrecka, J., Jawhari, A., Bruckner, F., Cr amer, P. & Michaelis, J. A Nano‐Positioning System for Macromolecular Structural Analysis. Nat. Meth. 5, 965‐971 (2008)

    2008

  3. [12]

    J., Ber ger, S., Restle, T., Goody, R

    Kalinin, S., Peulen, T., Sindbert, S., Rothwell, P. J., Ber ger, S., Restle, T., Goody, R. S., Gohlke, H. & Seidel, C. A. M . A toolkit and benchmark study for FRET‐restrained high‐precision structural modeling. Nat. Meth. 9, 1218‐1227 (2012)

    2012

  4. [13]

    & Gohlke, H

    Ahmed, A., Rippmann, F., Barnickel, G. & Gohlke, H. A norma l mode‐based geometric simulation approach for exploring biologically relevant conformational transitions in proteins. J. Chem. Inf. Model. 51, 1604‐1622 (2011)

    2011

  5. [14]

    N., Li, W

    Kinch, L. N., Li, W. L., Mona styrskyy, B., Kryshtafovych, A. & Grishin, N. V. Evaluation of free modeling targets in CASP11 and ROLL. Proteins‐Structure Function and Bioinformatics 84, 51‐66 (2016)

    2016

  6. [15]

    Brunger, A. T. Free R‐Value ‐ a Novel Statistical Quantity for Assessing the Accuracy of Crystal‐Structures. Nature 355, 472‐475 (1992)

    1992

  7. [16]

    Okazaki, K. I. & Takada, S. Dynamic energy landscape view o f coupled binding and protein conformational change: Induced‐fit versus population‐shift mechanisms. Proc. Natl. Acad. Sci. USA 105, 11182‐11187 (2008)

    2008

  8. [17]

    Identification of specific interactions that dr ive ligand‐induced closure in five enzymes with classic domain movements

    Hayward, S. Identification of specific interactions that dr ive ligand‐induced closure in five enzymes with classic domain movements. J. Mol. Biol. 339, 1001‐1021 (2004)

    2004

  9. [18]

    N., Kovacs, J

    Cavasotto, C. N., Kovacs, J. A. & Abagyan, R. A. Representi ng receptor flexibility in ligand docking through relevant normal modes. J. Am. Chem. Soc. 127, 9632‐9640 (2005)

    2005

  10. [19]

    & Gohlke, H

    Ahmed, A. & Gohlke, H. Multiscale modeling of macromolecular conformational changes combining concepts from rigidity and elastic network theory. Proteins‐Structure Function and Bioinformatics 63, 1038‐1051 (2006)

    2006

  11. [20]

    & Sanejouand, Y

    Tama, F. & Sanejouand, Y. H. Conformational change of prote ins arising from normal mode calculations. Protein Eng. 14, 1‐6 (2001)

    2001

  12. [21]

    R., Dimura, M., Koberling, F., Kühnemuth, R

    Sanabria, H., Rodnin, D., Hemmen, K., Peulen, T., Felekyan, S., Fleissner, M. R., Dimura, M., Koberling, F., Kühnemuth, R . , H u b b e l l , W . L . , G o h l k e , H . & S e i d e l , C . A . R e s o l v i n g d y n a m ics and function of transient states in single enzyme m...

    2018

  13. [22]

    & Wishart, D

    Berjanskii, M. & Wishart, D. S. NMR: prediction of protein flexibility. Nature Protocols 1, 683‐688 (2006)

    2006

  14. [23]

    Muller, C. W. & Schulz, G. E. Structure of the complex betw een adenylate kinase from Escherichia coli and the inhibitor Ap5A refined at 1.9 A resolution. A model for a catalytic transition state. J. Mol. Biol. 224, 159‐177 (1992)

    1992

  15. [24]

    A., Vincent, M

    McPhalen, C. A., Vincent, M. G. & Jansonius, J. N. X‐ray st ructure refinement and comparison of three forms of mitochondrial aspartate aminotransferase. J. Mol. Biol. 225, 495‐517 (1992)

    1992

  16. [25]

    Kuboniwa, H., Tjandra, N., Grzesiek, S., Ren, H., Klee, C. B. & Bax, A. Solution structure of calcium‐free calmodulin. Nat. Struct. Biol. 2, 768‐776 (1995)

    1995

  17. [26]

    B., Kurihara, H., Orita, M., Shibanuma, T., Furuya, T

    Osawa, M., Tokumitsu, H., S windells, M. B., Kurihara, H., Orita, M., Shibanuma, T., Furuya, T. & Ikura, M. A novel target recognition revealed by calmodulin in complex with Ca2+‐calmodulin‐dependent kinase kinase. Nat. Struct. Biol. 6, 819‐ 824 (1999). 17

    1999

  18. [27]

    H., Pandit, J., Kang, C

    Oh, B. H., Pandit, J., Kang, C. H., Nikaido, K., Gokcen, S. , Ames, G. F. & Kim, S. H. Three‐dimensional structures of the periplasmic lysine/arginine/ornithine‐binding protein with and without a ligand. J . B i o l . C h e m . 268, 11348‐11355 (1993)

    1993

  19. [28]

    Byrnes, L. J. & Sondermann, H. Structural basis for the nucleotide‐dependent dimerization of the large G protein atlastin‐1/SPG3A. Proc. Natl. Acad. Sci. USA 108, 2216‐2221 (2011)

    2011

  20. [29]

    W., Schlauderer, G

    Muller, C. W., Schlauderer, G. J., Reinstein, J. & Schulz, G. E. Adenylate kinase motions during catalysis: an energetic counterweight balancing substrate binding. Structure 4, 147‐156 (1996)

    1996

  21. [30]

    J., Singh, A., Szeto, K., Benvin, N

    Byrnes, L. J., Singh, A., Szeto, K., Benvin, N. M., O'Donne ll, J. P., Zipfel, W. R. & Sondermann, H. Structural basis for conformational switching and GTP loading of the large G protein atlastin. EMBO J. 32, 369‐384 (2013)

    2013

  22. [31]

    & Wilson, M

    Prahlad, J. & Wilson, M. A. Crystal structure of E. coli YaaA, a member of the DUF328/UPF0246 family (2015)

    2015

  23. [32]

    A., Vincent, M

    McPhalen, C. A., Vincent, M. G., Picot, D., Jansonius, J. N ., Lesk, A. M. & Chothia, C. Domain closure in mitochondrial aspartate aminotransferase. J. Mol. Biol. 227, 197‐213 (1992)

    1992

  24. [33]

    fastcluster: Fast hierarchical, agglomerative clustering routines for R and Python

    Mullner, D. fastcluster: Fast hierarchical, agglomerative clustering routines for R and Python. Journal of Statistical Software 53, 1‐18 (2013)

    2013

  25. [34]

    & Ding, C

    Peng, H., Long, F. & Ding, C. Feature selection based on mu tual information: criteria of max‐dependency, max‐ relevance, and min‐redundancy. IEEE Trans Pattern Anal Mach Intell 27, 1226‐1238 (2005)

    2005

  26. [35]

    J., Rader, A

    Jacobs, D. J., Rader, A. J., Kuhn, L. A. & Thorpe, M. F. Pr otein flexibility predictions using graph theory. Proteins 44, 150‐ 165 (2001)

    2001

  27. [36]

    Hermans, S. M. A., Pfleger, C., Nutschel, C., Hanke, C. A. & Gohlke, H. Rigidity theory for biomolecules: concepts, software, and applications. Wiley Interdisciplinary Reviews‐Computational Molecular Science 7 (2017)

    2017

  28. [37]

    & Gohlke, H

    Ahmed, A. & Gohlke, H. Multiscale modeling of macromolecular conformational changes combining concepts from rigidity and elastic network theory. Proteins 63, 1038‐1051 (2006)

    2006

  29. [38]

    O., Opanasyuk, O

    Peulen, T. O., Opanasyuk, O. & Seidel, C. A. M. Combining G raphical and Analytical Methods with Molecular Simulations To Analyze Time‐Resolved FRET Measurements of Labeled Macromolecules Accurately. J. Phys. Chem. B 121, 8211‐8241 (2017)

    2017

  30. [39]

    & Weiner, J

    Gao, J. & Weiner, J. H. Range of Validity of the Entropic S pring Concept in Polymer Melt Relaxation. Macromolecules 25, 3462‐3467 (1992)

    1992

  31. [40]

    AMBER 2017 (University of Ca lifornia, San Francisco, 2017)

    2017

  32. [41]

    A., Cheatham, T

    Case, D. A., Cheatham, T. E., Darden, T., Gohlke, H., Luo, R., Merz, K. M., Onufriev, A., Simmerling, C., Wang, B. & Woods, R. J. The Amber biomolecular simulation programs. J. Comput. Chem. 26, 1668‐1688 (2005)

    2005

  33. [42]

    L., Chandrasekhar, J., Madura, J

    Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W. & Klein, M. L. Comparison of Simple Potential Functions for Simulating Liquid Water. J. Chem. Phys. 79, 926‐935 (1983)

    1983

  34. [43]

    A., Martinez, C., Kasavajhala, K., Wickstrom, L., Hauser, K

    Maier, J. A., Martinez, C., Kasavajhala, K., Wickstrom, L., Hauser, K. E. & Simmerling, C. ff14SB: Improving the Accuracy of Protein Side Chain and Backbone Parameters from ff99SB. J. Chem. Theory Comput. 11, 3696‐3713 (2015)

    2015

  35. [44]

    W., Poole, D., Le Grand, S

    Salomon‐Ferrer, R., Gotz, A. W., Poole, D., Le Grand, S. & Walker, R. C. Routine Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 2. Explicit Solvent Particle Mesh Ewald. J. Chem. Theory Comput. 9, 3878‐3888 (2013)

    2013

  36. [45]

    P., Ciccotti, G

    Ryckaert, J. P., Ciccotti, G. & Berendsen, H. J. C. Numeric al‐Integration of Cartesian Equations of Motion of a System with Constraints ‐ Molecular‐Dynamics of N‐Alkanes. J. Comput. Phys. 23, 327‐341 (1977)

    1977

  37. [46]

    & Pedersen, L

    Darden, T., York, D. & Pedersen, L. Particle Mesh Ewald ‐ a n N.Log(N) Method for Ewald Sums in Large Systems. J. Chem. Phys. 98, 10089‐10092 (1993)

    1993

  38. [47]

    W., Le Grand, S., Walker, R

    Hopkins, C. W., Le Grand, S., Walker, R. C. & Roitberg, A. E. Long‐Time‐Step Molecular Dynamics through Hydrogen Mass Repartitioning. J. Chem. Theory Comput. 11, 1864‐1874 (2015)

    2015

  39. [48]

    Berendsen, H. J. C., Postma, J. P. M., Vangunsteren, W. F., Dinola, A. & Haak, J. R. Molecular‐Dynamics with Coupling to an External Bath. J. Chem. Phys. 81, 3684‐3690 (1984)

    1984

  40. [49]

    R., Brustad, E

    Fleissner, M. R., Brustad, E. M., Kalai, T., Altenbach, C., Cascio, D., Peters, F. B., Hideg, K., Peuker, S., Schultz, P. G. & Hubbell, W. L. Site‐directed spin labeling of a genetically encoded unnatural amino acid. Proc. Natl. Acad. Sci. USA 106, 21637‐21642 (2009)

    2009

  41. [50]

    Lemke, E. A. & Schultz, C. Principles for designing fluores cent sensors and reporters. Nat. Chem. Biol. 7, 480‐483 (2011)

    2011

  42. [51]

    M., Lemke, E

    Brustad, E. M., Lemke, E. A., Schultz, P. G. & Deniz, A. A. A General and Efficient Method for the Site‐Specific Dual‐ Labeling of Proteins for Single Molecule Fluorescence Resonance E n e r g y T r a n s f e r . J. Am. Chem. Soc. 130, 17664‐+ (2008)

    2008

  43. [52]

    O., Opanasyuk, O

    Peulen, T. O., Opanasyuk, O. & Seidel, C. A. M. Combining G raphical and Analytical Methods with Molecular Simulations To Analyze Time‐Resolved FRET Measurements of Labeled Macromolecules Accurately. J. Phys. Chem. B 121, 8211‐8241 (2017). 18

    2017

  44. [53]

    & Seidel, C

    Sindbert, S., Kalin in, S., Hien, N., Kienzler, A., Clima, L., Bannwarth, W., Appel, B., Muller, S. & Seidel, C. A. M. Accurate Distance Determination of Nucleic Acids via Förster Resonance Energy Transfer: Implications of Dye Linker Length and Rigidity. J. Am. Chem. Soc. 133,...

    2011

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.