Browse
Submit Data
Databases
API
Help

Dataset Information

0 Views

0 Connections

0 Citations

0 Reanalyses

0 Downloads

Omics score: 0

ManyFold: an efficient and flexible library for training and validating protein folding models.

ABSTRACT:

Summary

ManyFold is a flexible library for protein structure prediction with deep learning that (i) supports models that use both multiple sequence alignments (MSAs) and protein language model (pLM) embedding as inputs, (ii) allows inference of existing models (AlphaFold and OpenFold), (iii) is fully trainable, allowing for both fine-tuning and the training of new models from scratch and (iv) is written in Jax to support efficient batched operation in distributed settings. A proof-of-concept pLM-based model, pLMFold, is trained from scratch to obtain reasonable results with reduced computational overheads in comparison to AlphaFold.

Availability and implementation

The source code for ManyFold, the validation dataset and a small sample of training data are available at https://github.com/instadeepai/manyfold.

Supplementary information

Supplementary data are available at Bioinformatics online.

SUBMITTER: Villegas-Morcillo A

PROVIDER: S-EPMC9825755 | biostudies-literature | 2023 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Json Xml

Publications

ManyFold: an efficient and flexible library for training and validating protein folding models.

Villegas-Morcillo Amelia A Robinson Louis L Flajolet Arthur A Barrett Thomas D TD

Bioinformatics (Oxford, England) 20230101 1

<h4>Summary</h4>ManyFold is a flexible library for protein structure prediction with deep learning that (i) supports models that use both multiple sequence alignments (MSAs) and protein language model (pLM) embedding as inputs, (ii) allows inference of existing models (AlphaFold and OpenFold), (iii) is fully trainable, allowing for both fine-tuning and the training of new models from scratch and (iv) is written in Jax to support efficient batched operation in distributed settings. A proof-of-con ...[more]

PMID: 36495196

Similar Datasets

Highly Efficient Computation of the Basal kon using Direct Simulation of Protein-Protein Association with Flexible Molecular Models.

Project description:An essential baseline for determining the extent to which electrostatic interactions enhance the kinetics of protein-protein association is the "basal" kon, which is the rate constant for association in the absence of electrostatic interactions. However, since such association events are beyond the milliseconds time scale, it has not been practical to compute the basal kon by directly simulating the association with flexible models. Here, we computed the basal kon for barnase and barstar, two of the most rapidly associating proteins, using highly efficient, flexible molecular simulations. These simulations involved (a) pseudoatomic protein models that reproduce the molecular shapes, electrostatic, and diffusion properties of all-atom models, and (b) application of the weighted ensemble path sampling strategy, which enhanced the efficiency of generating association events by >130-fold. We also examined the extent to which the computed basal kon is affected by inclusion of intermolecular hydrodynamic interactions in the simulations.

| S-EPMC4716576 | biostudies-literature

Efficient flexible backbone protein-protein docking for challenging targets.

Project description:MotivationBinding-induced conformational changes challenge current computational docking algorithms by exponentially increasing the conformational space to be explored. To restrict this search to relevant space, some computational docking algorithms exploit the inherent flexibility of the protein monomers to simulate conformational selection from pre-generated ensembles. As the ensemble size expands with increased flexibility, these methods struggle with efficiency and high false positive rates.ResultsHere, we develop and benchmark RosettaDock 4.0, which efficiently samples large conformational ensembles of flexible proteins and docks them using a novel, six-dimensional, coarse-grained score function. A strong discriminative ability allows an eight-fold higher enrichment of near-native candidate structures in the coarse-grained phase compared to RosettaDock 3.2. It adaptively samples 100 conformations each of the ligand and the receptor backbone while increasing computational time by only 20-80%. In local docking of a benchmark set of 88 proteins of varying degrees of flexibility, the expected success rate (defined as cases with ≥50% chance of achieving 3 near-native structures in the 5 top-ranked ones) for blind predictions after resampling is 77% for rigid complexes, 49% for moderately flexible complexes and 31% for highly flexible complexes. These success rates on flexible complexes are a substantial step forward from all existing methods. Additionally, for highly flexible proteins, we demonstrate that when a suitable conformer generation method exists, the method successfully docks the complex.Availability and implementationAs a part of the Rosetta software suite, RosettaDock 4.0 is available at https://www.rosettacommons.org to all non-commercial users for free and to commercial users for a fee.Supplementary informationSupplementary data are available at Bioinformatics online.

| S-EPMC6184633 | biostudies-literature

Flexible and efficient simulation-based inference for models of decision-making.

Project description:Inferring parameters of computational models that capture experimental data are a central task in cognitive neuroscience. Bayesian statistical inference methods usually require the ability to evaluate the likelihood of the model-however, for many models of interest in cognitive neuroscience, the associated likelihoods cannot be computed efficiently. Simulation-based inference (SBI) offers a solution to this problem by only requiring access to simulations produced by the model. Previously, Fengler et al. introduced likelihood approximation networks (LANs, Fengler et al., 2021) which make it possible to apply SBI to models of decision-making, but require billions of simulations for training. Here, we provide a new SBI method that is substantially more simulation efficient. Our approach, mixed neural likelihood estimation (MNLE), trains neural density estimators on model simulations to emulate the simulator, and is designed to capture both the continuous (e.g., reaction times) and discrete (choices) data of decision-making models. The likelihoods of the emulator can then be used to perform Bayesian parameter inference on experimental data using standard approximate inference methods like Markov Chain Monte Carlo sampling. We demonstrate MNLE on two variants of the drift-diffusion model and show that it is substantially more efficient than LANs: MNLE achieves similar likelihood accuracy with six orders of magnitude fewer training simulations, and is significantly more accurate than LANs when both are trained with the same budget. Our approach enables researchers to perform SBI on custom-tailored models of decision-making, leading to fast iteration of model design for scientific discovery.

| S-EPMC9374439 | biostudies-literature

Exploring Protein-Nanoparticle Interactions with Coarse-Grained Protein Folding Models.

Project description:Understanding the fundamental biophysics behind protein-nanoparticle (NP) interactions is essential for the design and engineering bio-NP systems. The authors describe the development of a coarse-grained protein-NP model that utilizes a structure centric protein model. A key feature of the protein-NP model is the quantitative inclusion of the hydrophobic character of residues in the protein and their interactions with the NP surface. In addition, the curvature of the NP is taken into account, capturing the protein behavior on NPs of different size. The authors evaluate this model by comparison with experimental results for structure and adsorption of a model protein interacting with an NP. It is demonstrated that the simulation results recapitulate the structure of the small ?/? protein GB1 on the NP for data from circular dichroism and fluorescence spectroscopy. In addition, the calculated protein adsorption free energy agrees well with the experimental value. The authors predict the dependence of protein folding on the NP size, surface chemistry, and temperature. The model has the potential to guide NP design efforts by predicting protein behavior on NP surfaces with various chemical properties and curvatures.

| S-EPMC5551056 | biostudies-literature

Visualization of protein folding funnels in lattice models.

Project description:Protein folding occurs in a very high dimensional phase space with an exponentially large number of states, and according to the energy landscape theory it exhibits a topology resembling a funnel. In this statistical approach, the folding mechanism is unveiled by describing the local minima in an effective one-dimensional representation. Other approaches based on potential energy landscapes address the hierarchical structure of local energy minima through disconnectivity graphs. In this paper, we introduce a metric to describe the distance between any two conformations, which also allows us to go beyond the one-dimensional representation and visualize the folding funnel in 2D and 3D. In this way it is possible to assess the folding process in detail, e.g., by identifying the connectivity between conformations and establishing the paths to reach the native state, in addition to regions where trapping may occur. Unlike the disconnectivity maps method, which is based on the kinetic connections between states, our methodology is based on structural similarities inferred from the new metric. The method was developed in a 27-mer protein lattice model, folded into a 3×3×3 cube. Five sequences were studied and distinct funnels were generated in an analysis restricted to conformations from the transition-state to the native configuration. Consistent with the expected results from the energy landscape theory, folding routes can be visualized to probe different regions of the phase space, as well as determine the difficulty in folding of the distinct sequences. Changes in the landscape due to mutations were visualized, with the comparison between wild and mutated local minima in a single map, which serves to identify different trapping regions. The extension of this approach to more realistic models and its use in combination with other approaches are discussed.

| S-EPMC4091862 | biostudies-literature

FragBuilder: an efficient Python library to setup quantum chemistry calculations on peptides models.

Project description:We present a powerful Python library to quickly and efficiently generate realistic peptide model structures. The library makes it possible to quickly set up quantum mechanical calculations on model peptide structures. It is possible to manually specify a specific conformation of the peptide. Additionally the library also offers sampling of backbone conformations and side chain rotamer conformations from continuous distributions. The generated peptides can then be geometry optimized by the MMFF94 molecular mechanics force field via convenient functions inside the library. Finally, it is possible to output the resulting structures directly to files in a variety of useful formats, such as XYZ or PDB formats, or directly as input files for a quantum chemistry program. FragBuilder is freely available at https://github.com/jensengroup/fragbuilder/ under the terms of the BSD open source license.

| S-EPMC3961104 | biostudies-literature

Understanding protein domain-swapping using structure-based models of protein folding.

Project description:In domain-swapping, two or more identical protein monomers exchange structural elements and fold into dimers or multimers whose units are structurally similar to the original monomer. Domain-swapping is of biotechnological interest because inhibiting domain-swapping can reduce disease-causing fibrillar protein aggregation. To achieve such inhibition, it is important to understand both the energetics that stabilize the domain-swapped structure and the protein dynamics that enable the swapping. Structure-based models (SBMs) encode the folded structure of the protein in their potential energy functions. SBMs have been successfully used to understand diverse aspects of monomer folding. Symmetrized SBMs model interactions between two identical protein chains using only intra-monomer interactions. Molecular dynamics simulations of such symmetrized SBMs have been used to correctly predict the domain-swapped structure and to understand the mechanism of domain-swapping. Here, we review such models and illustrate that monomer topology determines key aspects of domain-swapping. However, in some proteins, specifics of local energetic interactions modulate domain-swapping and these need to be added to the symmetrized SBMs. We then summarize some general principles of the mechanism of domain-swapping that emerge from the symmetrized SBM simulations. Finally, using our own results, we explore how symmetrized SBMs could be used to design domain-swapping in proteins.

| S-EPMC7127520 | biostudies-literature

Efficient conformational space exploration in ab initio protein folding simulation.

Project description:Ab initio protein folding simulation largely depends on knowledge-based energy functions that are derived from known protein structures using statistical methods. These knowledge-based energy functions provide us with a good approximation of real protein energetics. However, these energy functions are not very informative for search algorithms and fail to distinguish the types of amino acid interactions that contribute largely to the energy function from those that do not. As a result, search algorithms frequently get trapped into the local minima. On the other hand, the hydrophobic-polar (HP) model considers hydrophobic interactions only. The simplified nature of HP energy function makes it limited only to a low-resolution model. In this paper, we present a strategy to derive a non-uniform scaled version of the real 20×20 pairwise energy function. The non-uniform scaling helps tackle the difficulty faced by a real energy function, whereas the integration of 20×20 pairwise information overcomes the limitations faced by the HP energy function. Here, we have applied a derived energy function with a genetic algorithm on discrete lattices. On a standard set of benchmark protein sequences, our approach significantly outperforms the state-of-the-art methods for similar models. Our approach has been able to explore regions of the conformational space which all the previous methods have failed to explore. Effectiveness of the derived energy function is presented by showing qualitative differences and similarities of the sampled structures to the native structures. Number of objective function evaluation in a single run of the algorithm is used as a comparison metric to demonstrate efficiency.

| S-EPMC4555859 | biostudies-literature

Dynamic folding pathway models of the Trp-cage protein.

Project description:Using action-derived molecular dynamics (ADMD), we study the dynamic folding pathway models of the Trp-cage protein by providing its sequential conformational changes from its initial disordered structure to the final native structure at atomic details. We find that the numbers of native contacts and native hydrogen bonds are highly correlated, implying that the native structure of Trp-cage is achieved through the concurrent formations of native contacts and native hydrogen bonds. In early stage, an unfolded state appears with partially formed native contacts (~40%) and native hydrogen bonds (~30%). Afterward, the folding is initiated by the contact of the side chain of Tyr3 with that of Trp6, together with the formation of the N-terminal ? -helix. Then, the C-terminal polyproline structure docks onto the Trp6 and Tyr3 rings, resulting in the formations of the hydrophobic core of Trp-cage and its near-native state. Finally, the slow adjustment processes of the near-native states into the native structure are dominant in later stage. The ADMD results are in agreement with those of the experimental folding studies on Trp-cage and consistent with most of other computational studies.

| S-EPMC3707288 | biostudies-literature

The McNorm library: creating and validating a new library of emotionally expressive whole body dance movements.

Project description:The ability to exchange affective cues with others plays a key role in our ability to create and maintain meaningful social relationships. We express our emotions through a variety of socially salient cues, including facial expressions, the voice, and body movement. While significant advances have been made in our understanding of verbal and facial communication, to date, understanding of the role played by human body movement in our social interactions remains incomplete. To this end, here we describe the creation and validation of a new set of emotionally expressive whole-body dance movement stimuli, named the Motion Capture Norming (McNorm) Library, which was designed to reconcile a number of limitations associated with previous movement stimuli. This library comprises a series of point-light representations of a dancer's movements, which were performed to communicate to observers neutrality, happiness, sadness, anger, and fear. Based on results from two validation experiments, participants could reliably discriminate the intended emotion expressed in the clips in this stimulus set, with accuracy rates up to 60% (chance = 20%). We further explored the impact of dance experience and trait empathy on emotion recognition and found that neither significantly impacted emotion discrimination. As all materials for presenting and analysing this movement library are openly available, we hope this resource will aid other researchers in further exploration of affective communication expressed by human bodily movement.

| S-EPMC8985749 | biostudies-literature

OmicsDI is part of the ELIXIR infrastructure

OmicsDI is an Elixir interoperability service. Learn more ›

Tweets

OmicsDI Databases

PRIDE
PeptideAtlas
MassIVE
JPOST Repository
Physiome Model Repository

EGA
EVA
ENA
LINCS
PAXDB
Cell Collective

MetaboLights
Metabolomics Workbench
MetabolomeExpress
GNPS
BioModels
FAIRDOMHub

ArrayExpress
dbGaP
ExpressionAtlas
GEO
NODE

Information

Databases
Help
API
Contact us
Code on GitHub
Terms of use
Submit Data