Energy-based models for atomic-resolution protein conformations

By Yilun Du, Joshua Meier, Jerry Ma, Rob Fergus, Alexander Rives. 2020

61 citations recorded by the source corpus.

We propose an energy-based model (EBM) of protein conformations that operates at atomic scale. The model is trained solely on crystallized protein data. By contrast, existing approaches for scoring conformations use energy functions that incorporate knowledge of physical principles and features that are the complex product of several decades of research and tuning. To evaluate our model, we benchmark on the rotamer recovery task, a restricted problem setting used to evaluate energy functions for protein design. Our model achieves comparable performance to the Rosetta energy function, a state-of-the-art method widely used in protein structure prediction and design. An investigation of the model’s outputs and hidden representations find that it captures physicochemical properties relevant to protein energy.

Subfields: Materials Chemistry, Molecular Biology, Probabilistic & Bayesian Methods, Computational Biology & Bioinformatics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Materials Chemistry

In Molecular Biology

In Probabilistic & Bayesian Methods

In Computational Biology & Bioinformatics

Citations in the corpus

Rethinking Attention with Performers (2020, background): “…2018), document generation/summarization, time series prediction, generative modeling (e.g. image generation (Parmar et al., 2018)), music generation (Huang et al., 2019), and bioinformatics (Rives et al., 2019; Madani et al., 2020; Ingraham et al., 2019; Elnaggar et al., 2019; Du et al., 2020).”
Protein sequence design with a learned potential (2022, methodology): “3A, B); this performance is on par with top benchmarked methods such as Rosetta 43 and another learned method for scoring rotamers 31 .”
Masked Language Modeling for Proteins via Linearly Scalable Long-Context Transformers (2020, background): “image generation [37]), music generation [27], and analysis of biological sequences [40, 34, 28, 21, 20].”
SE(3)-Equivariant Energy-based Models for End-to-End Protein Folding (2021, methodology): “In [8], random rotations are used as data augmentation to encourage (but not guarantee) the rotational invariance of the model for protein rotamer recovery, e.g. determining the best side-chain conformation with backbone structures given.”
Improved Contrastive Divergence Training of Energy Based Models (2020, background): “Energy-Based Models (EBMs) have the ability to compose with other models at generation time (Hinton, 1999; Haarnoja et al., 2017; Du et al., 2020a).”
Learning the Stein Discrepancy for Training and Evaluating Energy-Based Models without Sampling (2020, methodology): “These improvements enabled new applications in domains such as protein structure prediction (Ingraham et al., 2019; Du et al., 2020), and provided beneﬁt to fundamental problems in machine learning such as adversarial robustness, calibration, out-of-distribution detection (Grathwohl et al., 2019;…”
Predicting Molecular Conformation via Dynamic Graph Score Matching (2021, background): “Despite that there are some machine learning-based methods specialized in protein sidechain structure prediction [10], they are built upon rotamer libraries [5], which incorporates a lot of domain knowledge.”
Deep Learning in Protein Structural Modeling and Design (2020, methodology): “Du et al.(170) adopted an energy-based model (EBM)(171) to recover rotamers for backbone structures.”
Single Layers of Attention Suffice to Predict Protein Contacts (2020, background): “When protein structure is known, Ingraham et al. (2019) and Du et al. (2019) have used sparse attention based on the physical structure of the protein to directly encode priors about interaction.”