DNCON2: improved protein contact prediction using two-level deep convolutional neural networks

By Badri Adhikari, Jie Hou, Jianlin Cheng. 2017

148 citations recorded by the source corpus.

Motivation Significant improvements in the prediction of protein residue-residue contacts are observed in the recent years. These contacts, predicted using a variety of coevolution-based and machine learning methods, are the key contributors to the recent progress in ab initio protein structure prediction, as demonstrated in the recent CASP experiments. Continuing the development of new methods to reliably predict contact maps is essential to further improve ab initio structure prediction. Results In this paper we discuss DNCON2, an improved protein contact map predictor based on two-level deep convolutional neural networks. It consists of six convolutional neural networks – the first five predict contacts at 6, 7.5, 8, 8.5, and 10 Å distance thresholds, and the last one uses these five predictions as additional features to predict final contact maps. On the free-modeling datasets in CASP10, 11, and 12 experiments, DNCON2 achieves mean precisions of 35%, 50%, and 53.4%, respectively, higher than 30.6% by MetaPSICOV on CASP10 dataset, 34% by MetaPSICOV on CASP11 dataset, and 46.3% by Raptor-X on CASP12 dataset, when top L/5 long-range contacts are evaluated. We attribute the improved performance of DNCON2 to the inclusion of short- and medium-range contacts into training, two-level approach to prediction, use of the state-of-the-art optimization and activation functions, and a novel deep learning architecture that allows each filter in a convolutional layer to access all the input features of a protein of arbitrary length. Availability The web server of DNCON2 is at http://sysbio.rnet.missouri.edu/dncon2/ where training and testing datasets as well as the predictions for CASP10, 11, and 12 free-modeling datasets can also be downloaded. Its source code is available at https://github.com/multicom-toolbox/DNCON2/. Contact chengji@missouri.edu Supplementary information Supplementary data are available online.

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

Distance-based protein folding powered by deep learning (2018, methodology): “Cheng group, who proposed DBN, has switched to DCNN for contact prediction 13 .”
Deep convolutional neural networks for predicting the quality of single protein structural models (2019, background): “Especially, the application of deep convolutional neural network (CNN) 7,36,37 (e.g. 1DCNN for sequential data and 2DCNN for image-like inputs) has achieved the promising performance and becomes one of the best machine learning methods for solving bioinformatics problems 7,36,37 .”
rawMSA: End-to-end Deep Learning Makes Protein Sequence Profiles and Feature Extraction obsolete (2018, methodology): “At the same time, the Deep Learning revolution has proved to be a useful tool for better integrating the growing amount and complexity of input features [41–43].”
Assessing the accuracy of contact predictions in CASP13 (2019, background): “After 2010, when first papers on correlated mutations were published, a number of refinements took place, such as combining a set of nonoverlapping contact predictions in a consensus approach and especially the application of various supervised deep neural network based approaches,(42) further impro”
Ensembling multiple raw coevolutionary features with deep residual neural networks for contact-map prediction in CASP13 (2019, background): “Due to its significant importance to protein structure prediction, the problem of contact-map prediction has drawn an increasing amount of attention, and as a result, several new approaches have been proposed within the last decade.(10-21) Initial studies on protein contact prediction focused on the”
Deep learning methods in protein structure prediction (2020, methodology): “DNCON2 [117] CM Multi-stage CNN HHblits, jackHMMer”
Folding non-homologous proteins by coupling deep-learning contact maps with I-TASSER assembly simulations (2021, unknown): “, 2018), and DNCON2 (Adhikari et al., 2017), three co-evolution-based methods: GREMLIN (Kamisetty et al.”