FUpred: detecting protein domains through deep-learning-based contact map prediction

Abstract

Motivation Protein domains are subunits that can fold and function independently. Correct domain boundary assignment is thus a critical step toward accurate protein structure and function analyses. There is, however, no efficient algorithm available for accurate domain prediction from sequence. The problem is particularly challenging for proteins with discontinuous domains, which consist of domain segments that are separated along the sequence. Results We developed a new algorithm, FUpred, which predicts protein domain boundaries utilizing contact maps created by deep residual neural networks coupled with coevolutionary precision matrices. The core idea of the algorithm is to retrieve domain boundary locations by maximizing the number of intra-domain contacts, while minimizing the number of inter-domain contacts from the contact maps. FUpred was tested on a large-scale dataset consisting of 2549 proteins and generated correct single- and multi-domain classifications with a Matthew’s correlation coefficient of 0.799, which was 19.1% (or 5.3%) higher than the best machine learning (or threading)-based method. For proteins with discontinuous domains, the domain boundary detection and normalized domain overlapping scores of FUpred were 0.788 and 0.521, respectively, which were 17.3% and 23.8% higher than the best control method. The results demonstrate a new avenue to accurately detect domain composition from sequence alone, especially for discontinuous, multi-domain proteins. Availability and implementation https://zhanglab.ccmb.med.umich.edu/FUpred. Supplementary information Supplementary data are available at Bioinformatics online.

OUP accepted manuscript

“…They could be potentially identified using disorder predictors (since they are disordered), methods that predict domain boundaries (since some of the DFLs link domains), and flexibility predictors (since they are flexible). Therefore, we consider the latest protein domain predictor FUpred ( Zheng et al , 2020 ) and one of the latest flexibility predictors, PredyFlexy ( de Brevern et al , 2012 ). We also compare against a selection of popular disorder predictors including DISOPRED3 ( Jones and Cozzetto, 2015 ) and IUPred2A ( Meszaros et al , 2018 ), the latter in two of its versions that focus on the prediction of long and short IDRs.…”

Protein domain identification methods and online resources

“… 2019 https://github.com/yuexujiang/DeepDom [41] FuPred Predict protein domain boundaries using predicted contact maps generated by ANN. 2020 https://zhanglab.ccmb.med.umich.edu/FUpred [44] …”

“…Most of the above sequence-based methods do not consider discontinuous domain predictions, while about 18% of proteins in the current PDB library have at least one discontinuous domain. ThreaDomEX [43] and FUpred [44] are two methods that pay special attention to discontinuous domain detection. ThraeDomEX can detect a discontinuous domain mainly by incorporating DomEx [45] , which can assemble non-consecutive segments following multiple threading template alignments.…”

Progressive and accurate assembly of multi-domain protein structures from cryo-EM density maps

“…The size of these proteins runs from 144 to 1,664 residues with the number of domains ranging from 2 to 8. To emulate the common real-life scenarios where the domain structures of target proteins are unknown, we predict the domain boundaries from sequence by a deep-learning contact-based program FUpred 14 , with the individual domain structures modelled by I-TASSER. Figure 4a, FUpred did an acceptable job in domain boundary prediction and correctly predicted the number of domains in 37 out of the 51 test proteins.…”

“…1) to create accurate complex structure models for multi-domain proteins from cryo-EM density maps. The pipeline can start from either experimentally determined domain structures or amino acid sequences, where in the latter case the domain split and individual domain structure modeling are performed with FUpred 14 and I-TASSER 15 , respectively. To systematically examine the strength and weakness, DEMO-EM was tested on a large-scale benchmark dataset consisting of various numbers of continuous and discontinuous domains over synthesized and experimental density maps.…”