CryoFold: Determining protein structures and data-guided ensembles from cryo-EM density maps

Shekhar, Mrinal; Terashi, Genki; Gupta, Chitrak; Sarkar, Daipayan; Debussche, Gaspard; Sisco, Nicholas J.; Nguyen, Jonathan; Mondal, Arup; Vant, John; Fromme, Petra; Horn, Wade D. Van; Tajkhorshid, Emad; Dill, Ken A.; Pérez, Alberto; Singharoy, Abhishek

doi:10.1016/j.matt.2021.09.004

Cited by 32 publications

(37 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For assessing protein folding quality, MD simulations could be used to model the inherent flexibility of the protein near the native state (as a surrogate of stability) or better quantifying solvent-exposed hydrophobic patches prone to oligomerization. More recent MD schemes could also be used to perform protein structure prediction guided by external knowledge from the design ( MacCallum et al, 2015 ; Shekhar et al, 2021 ), cross-validating predictions from AF or RF.…”

Section: Discussionmentioning

confidence: 99%

Identifying well-folded de novo proteins in the new era of accurate structure prediction

Peñas-Utrilla

Marcos

2022

Front. Mol. Biosci.

View full text Add to dashboard Cite

Computational de novo protein design tailors proteins for target structures and oligomerisation states with high stability, which allows overcoming many limitations of natural proteins when redesigned for new functions. Despite significant advances in the field over the past decade, it remains challenging to predict sequences that will fold as stable monomers in solution or binders to a particular protein target; thereby requiring substantial experimental resources to identify proteins with the desired properties. To overcome this, here we leveraged the large amount of design data accumulated in the last decade, and the breakthrough in protein structure prediction from last year to investigate on improved ways of selecting promising designs before experimental testing. We collected de novo proteins from previous studies, 518 designed as monomers of different folds and 2112 as binders against the Botulinum neurotoxin, and analysed their structures with AlphaFold2, RoseTTAFold and fragment quality descriptors in combination with other properties related to surface interactions. These features showed high complementarity in rationalizing the experimental results, which allowed us to generate quite accurate machine learning models for predicting well-folded monomers and binders with a small set of descriptors. Cross-validating designs with varied orthogonal computational techniques should guide us for identifying design imperfections, rescuing designs and making more robust design selections before experimental testing.

show abstract

Section: Discussionmentioning

confidence: 99%

Identifying well-folded de novo proteins in the new era of accurate structure prediction

Peñas-Utrilla

Marcos

2022

Front. Mol. Biosci.

View full text Add to dashboard Cite

show abstract

“…Although the flexible fitting method can search the structures fitted to the AFM images, the method is prone to overfit in the case of sparse data like AFM images. The use of a Bayesian engine such as MELD [ 55 ] may work in this situation for finding physically reasonable structures as well as avoiding overfitting, as was done in CryoEM data analysis (CryoFold [ 56 ]). On the other hand, a variety of ensemble refinements and approaches, including the maximum entropy method [ 57 ] and a Bayesian method [ 58 ], would be useful in generating structural ensembles that match AFM images.…”

Section: Discussionmentioning

confidence: 99%

Development of hidden Markov modeling method for molecular orientations and structure estimation from high-speed atomic force microscopy time-series images

et al. 2022

View full text Add to dashboard Cite

High-speed atomic force microscopy (HS-AFM) is a powerful technique for capturing the time-resolved behavior of biomolecules. However, structural information in HS-AFM images is limited to the surface geometry of a sample molecule. Inferring latent three-dimensional structures from the surface geometry is thus important for getting more insights into conformational dynamics of a target biomolecule. Existing methods for estimating the structures are based on the rigid-body fitting of candidate structures to each frame of HS-AFM images. Here, we extend the existing frame-by-frame rigid-body fitting analysis to multiple frames to exploit orientational correlations of a sample molecule between H frames in HS-AFM data due to the interaction with the stage. In the method, we treat HS-AFM data as time-series data, and they are analyzed with the hidden Markov modeling. Using simulated HS-AFM images of the taste receptor type 1 as a test case, the proposed method shows a more robust estimation of molecular orientations than the frame-by-frame analysis. The method is applicable in integrative modeling of conformational dynamics using HS-AFM data.

show abstract

“…16 For proteins of molecular mass 500 kDa or bigger, composed of 5000 residues or more, a single CPU is expected to take 5000 years of wall-clock time for sampling the collective ensembles using either molecular dynamics (MD) or Monte Carlo (MC) simulations; 19 even the fastest GPUs of the day will not rescue this situation. Data-guided enhanced sampling methodologies, such as MELD 11 (integrated with NAMD via the recently completed CryoFold plugin 12 ) or backbone tracing methodologies such as MAINMAST 20 or analogous methods, 9 by themselves, either remain system-size limited, generating ensembles for only local regions within a map, or require further further refinements using conjugate gradient minimization or MD simulation schemes to determine ensemble models.…”

Section: Introductionmentioning

confidence: 99%

“…However, the conformational heterogeneity that contributes to the uncertainty of the the experimental data is lost. Biology often employs such conformational diversity in problems of allostery and recognition, motivating further the need to refine experimental knowledge against an ensemble of models, 12 rather than a single model interpretation. In this article we explore whether, it is possible to recover portions of the conformations lost in brute-force MDFF by running multiple replicas of MDFF in parallel with adaptive decision making based on map-model consistency parameters.…”

Section: Introductionmentioning

confidence: 99%

“…Integrative modeling is an area of rapid methodological developments, wherein, atom-resolved structure(s) of biological systems are determined by merging data from multiple experimental sources with physics 1-3 and informatics-guided approaches. 4 These elegant fitting, 1-3,5-10 learning 11 sand inferencing [12][13][14][15][16] methodologies have been successful in resolving a range of structures, starting with soluble and membrane proteins up to sub-cellular complex architectures. 17,18 Integrative models routinely make it to top positions at the EMDB and PDB competitions, serving a diverse cross-section of the Biophysics community.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Adaptive Ensemble Refinement of Protein Structures in High Resolution Electron Microscopy Density Maps with Radical Augmented Molecular Dynamics Flexible Fitting

Sarkar

Lee

Vant

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Recent advances in cryo-electron microscopy (cryo-EM) has enabled modeling macromolecular complexes that are essential components of life. The density maps obtained from cryo-EM experiments is often integrated with ab-initio, knowledge-driven or first principles-based computational methods to build, fit and refine protein structures inside the electron density maps. Going beyond a single stationary-structure determination scheme, it is becoming more common to interpret the experimental data with a set of multiple physical models all of which contributes to the average observation seen by the experiment. Hence, there is a need to decide on the quality of an ensemble of protein structures on-the-fly, while refining them against the density maps. In this work, we demonstrate such adaptive decision making capabilities during flexible fitting of biomolecules. Our solution uses RADICAL tools (RCT) and we test this new implementation in exascale high performance computing environment for two proteins, Adenylate Kinase (ADK) and Carbon Monoxide Dehydrogenase (CODH). Our results indicate that using multiple replicas in flexible fitting with adaptive decision making improves the overall quality of fit and model by 40 % improvement when compared against the traditional flexible fitting approach. These advances are agnostic to system-size and computing environments.

show abstract

CryoFold: Determining protein structures and data-guided ensembles from cryo-EM density maps

Cited by 32 publications

References 48 publications

Identifying well-folded de novo proteins in the new era of accurate structure prediction

Identifying well-folded de novo proteins in the new era of accurate structure prediction

Development of hidden Markov modeling method for molecular orientations and structure estimation from high-speed atomic force microscopy time-series images

Adaptive Ensemble Refinement of Protein Structures in High Resolution Electron Microscopy Density Maps with Radical Augmented Molecular Dynamics Flexible Fitting

Contact Info

Product

Resources

About