High precision in protein contact prediction using fully convolutional neural networks and minimal sequence features

By David T. Jones, Shaun M. Kandathil. 2018

169 citations recorded by the source corpus.

Subfields: Molecular Biology, Model Architectures, 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 Molecular Biology

In Model Architectures

In Computational Biology & Bioinformatics

Citations in the corpus

Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences (2019, methodology): “We use a dilated convolutional residual network similar to recent state-of-the-art methods for tertiary structure prediction (Xu, 2018; Jones & Kandathil, 2018; Senior et al., 2018).”
High-resolution de novo structure prediction from primary sequence (2022, methodology): “Similar to the ESM-1b model ( 12 ), the overall architecture of OmegaPLM is a self-attention model ( 24 ) in which each token is a single amino acid.”
Analysis of several key factors influencing deep learning-based inter-residue contact prediction (2019, background): “…architectures have been designed to integrate traditional sequence features with residue–residue coevolutionary scores to substantially improve the accuracy of contact prediction (Adhikari et al., 2018; Hanson et al., 2018; Jones and Kandathil, 2018; Skwark et al., 2014; Wang et al., 2017).”
DEEPCON: Protein Contact Prediction using Dilated Convolutional Neural Networks with Dropout (2019, methodology): “In [8] authors demonstrate that a basic CNN-based method can easily outperform another state-of-the-art meta-method based on basic neural networks.”
Prediction of interresidue contacts with DeepMetaPSICOV in CASP13 (2019, methodology): “To get the predicted scores for each residue pair, we average the values predicted for residue pairs (i,j) and (j,i) as in DeepCov.”
Transformer protein language models are unsupervised structure learners (2020, methodology): “Supervised contact prediction Recently, supervised methods using deep learning have resulted in breakthrough results in supervised contact prediction (Wang et al., 2017; Jones & Kandathil, 2018; Yang et al., 2019; Senior et al., 2020; Adhikari & Elofsson, 2020).”
ResPRE: high-accuracy protein contact prediction by coupling precision matrix with deep residual neural networks (2019, methodology): “DeepCov and DeepContact outperform these four programs with 3 and 4 contacts correctly predicted, respectively.”
Folding non-homologous proteins by coupling deep-learning contact maps with I-TASSER assembly simulations (2021, methodology): “In this study, we retrained NeBcon to improve its long-range contact prediction precision by using the naı¨ve Bayes classiﬁer (NBC) theorem to integrate eight state-of-the-art contact prediction methods, including four deep-learning-based methods: DeepPLM (Zheng et al., 2019a), DeepCov (Jones and Ka”
Assessing the accuracy of contact predictions in CASP13 (2019, background): “Despite the assumption that the input coupling matrix is random, 62 43 While more sequences have a positive impact on the accuracy of contact prediction in general, the deep residual neural networks have the ability to learn the underlying contact patterns from limited coevolutionary information;…”
DeepMSA: constructing deep multiple sequence alignment to improve contact prediction and fold-recognition for distant-homology proteins (2019, unknown): “In fact, for TripletRes and DeepCov, the MSA from DeepMSA yields slightly better contact precision compared to the MSA from DeepMSA Stage 3.”
Protein language model embeddings for fast, accurate, alignment-free protein structure prediction (2021, background): “Inputting co-evolution information, a narrow window/square around a pair of residues suffices to correctly infer contacts (Jones & Kandathil, 2018).”