PepFlow: direct conformational sampling from peptide energy landscapes through hypernetwork-conditioned diffusion

Abdin, Osama; Kim, Philip M.

doi:10.1101/2023.06.25.546443

“…However, we discovered that this approximation to work well in practice and discuss this choice in the Results section. When training, we always adopt batches containing protein conformations with the same number of residues to avoid padding strategies 23 .…”

Section: Sampling and Decodingmentioning

confidence: 99%

“…Deep generative models are a promising strategy for achieving this goal thanks to their capability of modeling complex probability distributions and fast sampling with one or few steps [19][20][21]. They have been applied to datasets of both folded [22] and intrinsically disordered proteins or peptides [23]. Most applications focus on replicating ensembles from simulations, but recently the DynamICE model [24] was used to integrate experimental data in the ensemble generation process of IDRs, highlighting the potential of generative models in integrative structural biology [9].…”

Section: Introductionmentioning

confidence: 99%

Transferable deep generative modeling of intrinsically disordered protein conformations

Janson,

Feig

2024

Preprint

View full text Add to dashboard Cite

Intrinsically disordered proteins have dynamic structures through which they play key biological roles. The elucidation of their conformational ensembles is a challenging problem requiring an integrated use of computational and experimental methods. Molecular simulations are a valuable computational strategy for constructing structural ensembles of disordered proteins but are highly resource-intensive. Recently, machine learning approaches based on deep generative models that learn from simulation data have emerged as an efficient alternative for generating structural ensembles. However, such methods currently suffer from limited transferability when modeling sequences and conformations absent in the training data. Here, we develop a novel generative model that achieves high levels of transferability for intrinsically disordered protein ensembles. The approach, named idpSAM, is a latent diffusion model based on transformer neural networks. It combines an autoencoder to learn a representation of protein geometry and a diffusion model to sample novel conformations in the encoded space. IdpSAM was trained on a large dataset of simulations of disordered protein regions performed with the ABSINTH implicit solvent model. Thanks to the expressiveness of its neural networks and its training stability, idpSAM faithfully captures 3D structural ensembles of test sequences with no similarity in the training set. Our study also demonstrates the potential for generating full conformational ensembles from datasets with limited sampling and underscores the importance of training set size for generalization. We believe that idpSAM represents a significant progress in transferable protein ensemble modeling through machine learning.

show abstract

“…We use different n systems and n frames values for the AE and the DDPM (the former is trained with only a part of the full dataset to speed up training) and different number of epochs. When training, we always adopt batches containing protein conformations with the same number of residues to avoid padding strategies 23 .…”

Section: Training Processmentioning

confidence: 99%

“…Deep generative models are a promising strategy for achieving this goal thanks to their capability of modeling complex probability distributions and fast sampling with one or few steps [19][20][21]. They have been applied to datasets of both folded [22] and intrinsically disordered proteins or peptides [23]. Most applications focus on replicating ensembles from simulations, but recently the DynamICE model [24] was used to integrate experimental data in the ensemble generation process of IDRs, highlighting the potential of generative models in integrative structural biology [9].…”

Section: Introductionmentioning

confidence: 99%

Transferable deep generative modeling of intrinsically disordered protein conformations

Janson,

Feig

2024

PLoS Comput Biol

4

0

View full text Add to dashboard Cite

Intrinsically disordered proteins have dynamic structures through which they play key biological roles. The elucidation of their conformational ensembles is a challenging problem requiring an integrated use of computational and experimental methods. Molecular simulations are a valuable computational strategy for constructing structural ensembles of disordered proteins but are highly resource-intensive. Recently, machine learning approaches based on deep generative models that learn from simulation data have emerged as an efficient alternative for generating structural ensembles. However, such methods currently suffer from limited transferability when modeling sequences and conformations absent in the training data. Here, we develop a novel generative model that achieves high levels of transferability for intrinsically disordered protein ensembles. The approach, named idpSAM, is a latent diffusion model based on transformer neural networks. It combines an autoencoder to learn a representation of protein geometry and a diffusion model to sample novel conformations in the encoded space. IdpSAM was trained on a large dataset of simulations of disordered protein regions performed with the ABSINTH implicit solvent model. Thanks to the expressiveness of its neural networks and its training stability, idpSAM faithfully captures 3D structural ensembles of test sequences with no similarity in the training set. Our study also demonstrates the potential for generating full conformational ensembles from datasets with limited sampling and underscores the importance of training set size for generalization. We believe that idpSAM represents a significant progress in transferable protein ensemble modeling through machine learning.

show abstract

“…RFdiffusion [Watson et al, 2023] integrated the diffusion model with a pre-trained RoseTTAFold to perform de novo protein design, motif scaffolding and binder design. While recent works have leveraged diffusion models for conformational distribution tasks by training on molecular force fields [Abdin and Kim, 2023, Zheng et al, 2023], limited work has been done on protein structure inpainting tasks where only a subset of the residues are of interest. By fixing the majority of residue positions, we sample the conformational distribution in the area of interest while avoiding incorrect global structure predictions.…”

Section: Introductionmentioning

confidence: 99%

FrameDiPT: SE(3) Diffusion Model for Protein Structure Inpainting

Zhang,

Leach,

Makkink

et al. 2023

Preprint

3

0

View full text Add to dashboard Cite

Protein structure prediction field has been revolutionised by deep learning with protein folding models such as AlphaFold 2 and ESMFold. These models enable rapidin silicoprediction and have been integrated intode novoprotein design and protein-protein interaction (PPI) prediction. However, biologically relevant features dependent on conformational distributions cannot be estimated with these models. Diffusion models, a novel class of generative models, have been developed to learn conformational distributions and applied tode novoprotein design. Limited work has been done on protein structure inpainting, where a masked section is recovered by simultaneously conditioning on its sequence and the rest of the structure. In this work, we proposeFrameDiffinPainTing (FrameDiPT), a generalised model for protein inpainting. This is important for T-cells given the hyper-variability of the complementarity determining region (CDR) loops. We evaluated the model on CDR loop design for T-cell receptors and achieved comparable prediction accuracy to ProteinGenerator and RFdiffusion with limited training data and learnable parameters. Different from deterministic structure prediction models, FrameDiPT captures the conformational distribution at different regions and binding states, highlighting a key advantage of generative models.

show abstract

PepFlow: direct conformational sampling from peptide energy landscapes through hypernetwork-conditioned diffusion

Cited by 4 publications

References 36 publications

Transferable deep generative modeling of intrinsically disordered protein conformations

Transferable deep generative modeling of intrinsically disordered protein conformations

Transferable deep generative modeling of intrinsically disordered protein conformations

FrameDiPT: SE(3) Diffusion Model for Protein Structure Inpainting

Contact Info

Product

Resources

About