Predicting protein residue-residue contacts using deep networks and boosting

By Jesse Eickholt, Jianlin Cheng. 2012

162 citations recorded by the source corpus.

Subfields: 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 Molecular Biology

1. Jinbo Xu
2. Jianlin Cheng
3. Sheng Wang
4. Badri Adhikari
5. Yang Zhang
6. Jie Hou
7. Arne Elofsson
8. Robin Pearce

In Computational Biology & Bioinformatics

1. Jinbo Xu
2. Sheng Wang
3. Badri Adhikari
4. Jianlin Cheng
5. Robin Pearce
6. Arne Elofsson
7. Yang Zhang
8. Jie Hou

Citations in the corpus

• Analysis of several key factors influencing deep learning-based inter-residue contact prediction (2019, background): “After deep learning contact prediction was first introduced in 2012 (Di Lena et al., 2012; Eickholt and Cheng, 2012), different deep learning architectures have been designed to integrate traditional sequence features with residue–residue coevolutionary scores to substantially improve the accuracy o”
• Distance-based protein folding powered by deep learning (2018, methodology): “Different from ResNet and DCA, both DBN and MetaPSICOV are local prediction methods because when predicting the relationship of one residue pair they ignore the status of the other pairs.”
• DeepSF: deep convolutional neural network for mapping protein sequences to folds (2017, background): “ccess in computer vision, speech recognition and natural language processing [20, 21]. The application of deep learning in bioinformatics has also gained the traction since 2012. Deep belief networks [22] were developed to predict protein residue-residue contacts. Recently a deep residual convolutio”
• A Deep Learning Network Approach to ab initio Protein Secondary Structure Prediction (2014, methodology): “This is a non-redundant set representative of the proteins contained in the PDB and originally curated in the construction of a residue-residue contact predictor [27].”
• Improved Contact Predictions Using the Recognition of Protein Like Contact Patterns (2014, methodology): “Unfortunately, we did not manage to run all queries through DNcon website, but preliminary data for about 100 proteins did not indicate that it would perform better than CMAPpro or PhyCMAP.”
• Deep learning for computational biology (2016, neutral): “Restricted Boltzmann machines can also be used for unsupervised pre-training of deep networks to subsequently train supervised models of protein secondary structures (Spencer et al , 2015), disordered protein regions (Eickholt & Cheng, 2013) or amino acid contacts (Eickholt & Cheng, 2012).”
• Protein contact prediction by integrating joint evolutionary coupling analysis and supervised learning (2013, background): “It may further improve contact prediction by integrating other supervised learning methods such as CMAPpro, NNcon and DNcon or even other EC methods into our GGL framework.”
• Analysis of distance-based protein structure prediction by deep learning in CASP13 (2019, methodology): “This is mainly because that Cheng's DBN method has almost the same performance as traditional machine learning methods such as Random Forests, as reported.(40,41) In CASP10, the DBNmethod outperformed Random Forests by only 1.”
• Analysis of deep learning methods for blind protein contact prediction in CASP12 (2017, background): “Both our self-test results and the CASP12 result indicate that deep learning is a good technique for protein contact prediction although previous deep learning methods such as DNCON [10] and CMAPpro [20] did not stand out in CASP11.”
• ConEVA: a toolbox for comprehensive assessment of protein contacts (2016, methodology): “Currently, for evaluating predicted contacts, the three most widely used evaluation measures are precision, coverage and distance distribution score (X d ) [3, 12, 14, 18 – 22].”