Learning Protein Structure with a Differentiable Simulator

By John Ingraham, Adam J. Riesselman, Christopher Sander, Debora Marks. 2018

161 citations recorded by the source corpus.

How researchers see this paper

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

Citations in the corpus

GeoDiff: a Geometric Diffusion Model for Molecular Conformation Generation (2022, background): “Ingraham et al. (2019) proposed neural networks to learn an energy simulator to infer the protein structures.”
Transformer protein language models are unsupervised structure learners (2020, background): “Other recent work with deep learning uses sequences or evolutionary features as inputs (AlQuraishi, 2018; Ingraham et al., 2019).”
Energy-based models for atomic-resolution protein conformations (2020, background): “Ingraham et al. (2018) proposed learning an energy function for protein folding by backpropagating through a differentiable simulator.”
Toward the solution of the protein structure prediction problem (2021, methodology): “NEMO uses a combination of 1D, 2D, and graph convolutions to predict interresidue distances, orientations, and dihedral angles and utilizes Langevin dynamics to generate models based on these predicted features.”
Deep learning methods in protein structure prediction (2020, methodology): “Thesametrendtowardsend-to-endsystemshasbeenattempted in the pipeline from processed homologous sequences to 3D structure, e.g. inNEMO [143], a differentiablesimulator,and RGN(Recurrent Geometrical Network) [144], an end-to-end differentiable learning of protein structure.”
OT-Flow: Fast and Accurate Continuous Normalizing Flows via Optimal Transport (2020, background): “They perform well in density estimation (Chen et al. 2018a; Grathwohl et al. 2019; Papamakarios et al. 2019) and inference (Ingraham et al. 2019; Papamakarios et al. 2019), especially in physics and computational chemistry (Noé et al. 2019; Brehmer et al. 2020).”
Gradients are Not All You Need (2021, background): “…network initialization and activation design [Yang and Schoenholz, 2017, Hayou et al., 2019], model based control [Parmas et al., 2018, Parmas], meta-learning [Metz et al., 2019], ﬂuid simulation [Ni and Wang, 2017, Kochkov et al., 2021], and learning protein structures [Ingraham et al., 2018].”
Learning the Stein Discrepancy for Training and Evaluating Energy-Based Models without Sampling (2020, methodology): “Recently, advances in generic gradient-based samplers (Welling & Teh, 2011) have allowed this gradient estimator to be used to train much more large-scale models (Du & Mordatch, 2019; Grathwohl et al., 2019; Ingraham et al., 2019).”
Learning Neural Generative Dynamics for Molecular Conformation Generation (2021, background): “Some other works (Lemke & Peter, 2019; AlQuraishi, 2019; Ingraham et al., 2019; No´e et al., 2019; Senior et al., 2020) also focus on learning models to directly generate 3D structure, but focus on the protein folding problem.”
JAX, M.D.: End-to-End Differentiable, Hardware Accelerated, Molecular Dynamics in Pure Python (2019, background): “…finance (Capriotti, 2010), atmospheric modelling (Charpentier & Ghemires, 2000; Carmichael et al., 1997), optimal control (Walther, 2007), physics engines (de Avila Belbute-Peres et al., 2018), protein modelling (Ingraham et al., 2018; AlQuraishi, 2019), and quantum circuits (Schuld et al., 2019).”
AlphaFold, Artificial Intelligence (AI), and Allostery (2022, methodology): “Deep learning has also been already exploited in structural modeling and design 60,194 − 196 and analysis 191 and linking these to function.”