Sixty-five years of the long march in protein secondary structure prediction: the final stretch?

By Yuedong Yang, Jianzhao Gao, Jihua Wang, Rhys Heffernan, Jack Hanson, Kuldip K. Paliwal, Yaoqi Zhou. 2016

235 citations recorded by the source corpus.

Abstract Protein secondary structure prediction began in 1951 when Pauling and Corey predicted helical and sheet conformations for protein polypeptide backbone even before the first protein structure was determined. Sixty-five years later, powerful new methods breathe new life into this field. The highest three-state accuracy without relying on structure templates is now at 82–84%, a number unthinkable just a few years ago. These improvements came from increasingly larger databases of protein sequences and structures for training, the use of template secondary structure information and more powerful deep learning techniques. As we are approaching to the theoretical limit of three-state prediction (88–90%), alternative to secondary structure prediction (prediction of backbone torsion angles and Cα-atom-based angles and torsion angles) not only has more room for further improvement but also allows direct prediction of three-dimensional fragment structures with constantly improved accuracy. About 20% of all 40-residue fragments in a database of 1199 non-redundant proteins have <6 Å root-mean-squared distance from the native conformations by SPIDER2. More powerful deep learning methods with improved capability of capturing long-range interactions begin to emerge as the next generation of techniques for secondary structure prediction. The time has come to finish off the final stretch of the long march towards protein secondary structure prediction.

Subfields: Materials Chemistry, Molecular Biology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Materials Chemistry

In Molecular Biology

Citations in the corpus

Modeling aspects of the language of life through transfer-learning protein sequences (2019, background): “All published values referred to the following three test sets (also referred to as validation set): TS115 [69]: 115 proteins from high-quality structures (< 3 Å) released after 2015 (and at most 30% PIDE to any protein of known structure in the PDB at the time); CB513 [70]: 513 non-redundant…”
NetSurfP-2.0: improved prediction of protein structural features by integrated deep learning (2018, methodology): “This effect is more pronounced for the datasets that have more disordered regions (CASP12 and TS115) and less so for the CB513 dataset.”
ProtTrans: Towards Cracking the Language of Life’s Code Through Self-Supervised Deep Learning and High Performance Computing (2020, positive): “Each of those has severe limitations (CASP12: too small, CB513 and TS115 redundant and outdated).”
Capturing non‐local interactions by long short‐term memory bidirectional recurrent neural networks for improving prediction of protein secondary structure, backbone angles, contact numbers and solvent accessibility (2017, methodology): “…area of research as the accuracy of three-state prediction increases slowly from<70% to 82–84% (Rost, 2001; Zhou and Faraggi, 2010; Heffernan et al., 2015; Wang et al., 2016; Mirabello and Pollastri, 2013; Yaseen and Li, 2014), approaching the theoretical limit in the range of 88–90% (Rost, 2001).”
ProtTrans: Towards Cracking the Language of Life’s Code Through Self-Supervised Learning (2021, unknown): “We also included other public test data sets, namely CB513 [47]), TS115 [48], and CASP12 [49].”
Modeling the language of life – Deep Learning Protein Sequences (2019, background): “For instance, the three state per-residue accuracy for secondary structure prediction improved by over eight percentage points through ELMo (Table 1: e.g. for CB513: Q3(DeepSeqVec)=76.9% vs. Q3(DeepBLOSUM65)=67.4%, i.e. 9.4 percentage points corresponding to 14% of 67.4), the per-residue MCC for pro”
rawMSA: End-to-end Deep Learning Makes Protein Sequence Profiles and Feature Extraction obsolete (2018, background): “One example of this comes from secondary structure prediction systems, which have recently been reported to predict at accuracies (Q3) of up to 84% [64], yet we have not been able to find a recent study where the reported accuracy is supported by a proper splitting of the training and”
NetSurfP-3.0: accurate and fast prediction of protein structural features by protein language models and deep learning (2022, methodology): “Its training and validation set consist of 10 337 and 500 non-redundant proteins, respectively, and the testing is performed on the three test sets TS115 (115 proteins) (17), CB513 (513 protein regions from 434 proteins) (18), and CASP12 (21 proteins) (19).”