End-to-end differentiable learning of protein structure

By Mohammed Alquraishi. 2018

353 citations recorded by the source corpus.

Predicting protein structure from sequence is a central challenge of biochemistry. Co‐evolution methods show promise, but an explicit sequence‐to‐structure map remains elusive. Advances in deep learning that replace complex, human‐designed pipelines with differentiable models optimized end‐to‐end suggest the potential benefits of similarly reformulating structure prediction. Here we report the first end‐to‐end differentiable model of protein structure. The model couples local and global protein structure via geometric units that optimize global geometry without violating local covalent chemistry. We test our model using two challenging tasks: predicting novel folds without co‐evolutionary data and predicting known folds without structural templates. In the first task the model achieves state‐of‐the‐art accuracy and in the second it comes within 1‐2Å; competing methods using co‐evolution and experimental templates have been refined over many years and it is likely that the differentiable approach has substantial room for further improvement, with applications ranging from drug discovery to protein design.

Subfields: Computational Theory and Mathematics, Molecular Biology, 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 Computational Theory and Mathematics

In Molecular Biology

In Computational Biology & Bioinformatics

Citations in the corpus

TANKBind: Trigonometry-Aware Neural NetworKs for Drug-Protein Binding Structure Prediction (2022, background): “…surge of interest in integrating geometric priors for representation learning in the domain of drug discovery Jumper et al. [2021], Baek et al. [2021], Jing et al. [2021], Ganea et al. [2021], Jin et al. [2021], Ingraham et al. [2019], AlQuraishi [2019], Schütt et al. [2017], Somnath et al. [2021].”
Highly accurate protein structure prediction with AlphaFold (2021, critical): “A few recent studies have been developed to predict the 3D coordinates directly 47–50 , but the accuracy of these approaches does not match traditional, hand-crafted structure prediction pipelines 51 .”
Improved protein structure prediction by deep learning irrespective of co-evolution information (2020, methodology): “To put our work into perspective, we compare it with a top server Robetta[24]in CASP13 and a popular deep learning method RGN (recurrent geometric networks) [25].”
Unified rational protein engineering with sequence-based deep representation learning (2019, methodology): “Instead of using the predicted structures of the RGN, we extracted the internal state learned by the model for each sequence + PSSM combination (3,200 dimensions corresponding to the last outputs of two bidirectional LSTMs, each comprised 800 units per direction).”
Protein structure prediction using multiple deep neural networks in the 13th Critical Assessment of Protein Structure Prediction (CASP13) (2019, background): “23 While this neural network is an end-to-end model of protein structure (an alternative formulation of which has been proposed in concurrent work 1 ), it was found to model local structure much better than long-range interactions, so was not effective for sampling conformations of entire domains.”
ProtTrans: Towards Cracking the Language of Life’s Code Through Self-Supervised Deep Learning and High Performance Computing (2020, background): “Recently, the leap of NLP through advanced LMs have successfully been generalized toward understanding the language of life through advanced LMs trained on proteins [16], [17], [18], [19], [20], [21], [22], [23], [24].”
GeoDiff: a Geometric Diffusion Model for Molecular Conformation Generation (2022, background): “AlQuraishi (2019) uses recurrent networks to model the amino acid sequences.”
ProteinNet: a standardized data set for machine learning of protein structure (2019, background): “d task. ProteinNet should help facilitate such applications, along with the development of end-to-end differentiable models of protein structure that can be directly fused to other learning problems (AlQuraishi, 2018). As the quality of protein structure prediction algorithms continues to improve, w”
Energy-based models for atomic-resolution protein conformations (2020, background): “AlQuraishi (2019) investigated predicting protein structure from sequence without using co-evolution.”
ProtTrans: Towards Cracking the Language of Life’s Code Through Self-Supervised Learning (2021, background): “The leap of NLP through advanced LMs has been successfully generalized toward understanding the language of life through advanced LMs trained on proteins [31], [32], [33], [34], [35], [36], [37], [38], [39].”
Generating Novel, Designable, and Diverse Protein Structures by Equivariantly Diffusing Oriented Residue Clouds (2023, background): “This approach sidesteps the embeddability problem of distance-based representations, but is overly reliant on reasoning over local geometry (AlQuraishi, 2019).”