The selection of papers in protein residue-residue contact prediction is biased.
The NeBcon paper (Baoji He, S. M. Mortuza, Yanting Wang, Hong-Bin Shen, Yang Zhang."NeBcon: protein contact map prediction using neural network training coupled with naïve Bayes classifiers." Bioinformatics.) does not use deep learning and is therefore irrelevant to this list.
The SPOT-Contact paper (Yaoqi Zhou. "Accurate Prediction of Protein Contact Maps by Coupling Residual Two-Dimensional Bidirectional Long Short-Term Memory with Convolutional Neural Networks." Bioinformatics. (19 June 2018).) and the DeepConPred paper (Dapeng Xiong, Jianyang Zeng, and Haipeng Gong. "A deep learning framework for improving long-range residue–residue contact prediction using a hierarchical strategy." Bioinformatics.) are far less influential than Jinbo Xu's earlier paper
Wang, Sheng, Siqi Sun, Zhen Li, Renyu Zhang, and Jinbo Xu. "Accurate de novo prediction of protein contact map by ultra-deep learning model." PLoS computational biology 13, no. 1 (2017): e1005324.
While this paper is not the first attempt that uses deep learning for protein contact prediction (both DNCON and PConSC2 proceed this work), it is the first work that shows deep learning can obtain significantly higher prediction accuracy in contact prediction than traditional machine learning such as shallow neural network. It is also the first paper to propose formulating contact prediction as pixel level labelling problem instead of image level labelling problem, and to use convolutional neural network (CNN) for this purpose. These ideas are so critical to the performance of contact prediction that almost all recent state-of-the-art deep learning based contact prediction programs, including SPOT-Contact, DNCON2, PConsC4, and DeepCov, follow the CNN based pixel level labelling idea proposed by this paper.
The other two listed paper from the same lab ("ComplexContact: a web server for inter-protein contact prediction using deep learning." and "Protein threading using residue co-variation and deep learning.") are either based on or heavily influenced by the above mentioned earlier work by Wang et al 2017.
The selection of papers in protein residue-residue contact prediction is biased.
The NeBcon paper (Baoji He, S. M. Mortuza, Yanting Wang, Hong-Bin Shen, Yang Zhang."NeBcon: protein contact map prediction using neural network training coupled with naïve Bayes classifiers." Bioinformatics.) does not use deep learning and is therefore irrelevant to this list.
The SPOT-Contact paper (Yaoqi Zhou. "Accurate Prediction of Protein Contact Maps by Coupling Residual Two-Dimensional Bidirectional Long Short-Term Memory with Convolutional Neural Networks." Bioinformatics. (19 June 2018).) and the DeepConPred paper (Dapeng Xiong, Jianyang Zeng, and Haipeng Gong. "A deep learning framework for improving long-range residue–residue contact prediction using a hierarchical strategy." Bioinformatics.) are far less influential than Jinbo Xu's earlier paper
Wang, Sheng, Siqi Sun, Zhen Li, Renyu Zhang, and Jinbo Xu. "Accurate de novo prediction of protein contact map by ultra-deep learning model." PLoS computational biology 13, no. 1 (2017): e1005324.
While this paper is not the first attempt that uses deep learning for protein contact prediction (both DNCON and PConSC2 proceed this work), it is the first work that shows deep learning can obtain significantly higher prediction accuracy in contact prediction than traditional machine learning such as shallow neural network. It is also the first paper to propose formulating contact prediction as pixel level labelling problem instead of image level labelling problem, and to use convolutional neural network (CNN) for this purpose. These ideas are so critical to the performance of contact prediction that almost all recent state-of-the-art deep learning based contact prediction programs, including SPOT-Contact, DNCON2, PConsC4, and DeepCov, follow the CNN based pixel level labelling idea proposed by this paper.
The other two listed paper from the same lab ("ComplexContact: a web server for inter-protein contact prediction using deep learning." and "Protein threading using residue co-variation and deep learning.") are either based on or heavily influenced by the above mentioned earlier work by Wang et al 2017.