A Unified Multitask Architecture for Predicting Local Protein Properties
By Yanjun Qi, Merja Oja, Jason Aaron Edward Weston, William Stafford Noble. 2012
64 citations recorded by the source corpus.
A variety of functionally important protein properties, such as secondary structure, transmembrane topology and solvent accessibility, can be encoded as a labeling of amino acids. Indeed, the prediction of such properties from the primary amino acid sequence is one of the core projects of computational biology. Accordingly, a panoply of approaches have been developed for predicting such properties; however, most such approaches focus on solving a single task at a time. Motivated by recent, successful work in natural language processing, we propose to use multitask learning to train a single, joint model that exploits the dependencies among these various labeling tasks. We describe a deep neural network architecture that, given a protein sequence, outputs a host of predicted local properties, including secondary structure, solvent accessibility, transmembrane topology, signal peptides and DNA-binding residues. The network is trained jointly on all these tasks in a supervised fashion, augmented with a novel form of semi-supervised learning in which the model is trained to distinguish between local patterns from natural and synthetic protein sequences. The task-independent architecture of the network obviates the need for task-specific feature engineering. We demonstrate that, for all of the tasks that we considered, our approach leads to statistically significant improvements in performance, relative to a single task neural network approach, and that the resulting model achieves state-of-the-art performance.
Subfields: 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 Molecular Biology
Citations in the corpus
- Opportunities and obstacles for deep learning in biology and medicine (2017, methodology): “Several groups [28,191,192] initiated the application of deep learning to protein secondary structure prediction but were unable to achieve significant improvement over the de facto standard method PSIPRED [193], which uses two shallow feedforward neural networks.”
- Deep Supervised and Convolutional Generative Stochastic Network for Protein Secondary Structure Prediction (2014, methodology): “We discretized solvent accessibility scores to absolute solvent accessibility and relative solvent accessibility following (Qi et al., 2012).”
- A Deep Learning Network Approach to ab initio Protein Secondary Structure Prediction (2014, methodology): “Additionally, a couple of deep learning protein structure predictors have been developed recently, including a multifaceted prediction tool that predicts several protein structural elements in tandem [36] and a predictor that utilizes global protein information [41].”
- Deep Semantic Protein Representation for Annotation, Discovery, and Engineering (2018, background): “Several groups have deployed machine learning methods to identify specific functions of proteins 15–19 , however there have been only limited attempts to build a single classifier that assigns multiple features to a given peptide sequence 20–23 .”
- Improving protein disorder prediction by deep bidirectional long short‐term memory recurrent neural networks (2016, background): “…contact prediction (Di Lena et al. , 2012; Eickholt and Cheng, 2012), secondary structure and solvent accessible surface-area prediction (Heffernan et al. , 2015, 2016; Paliwal et al. , 2015; Qi et al. , 2012) and protein disorder prediction (Eickholt and Cheng, 2013; Paliwal et al. , 2015).”
- Sixty-five years of the long march in protein secondary structure prediction: the final stretch? (2016, methodology): “SPIDER2 was not the first method that used deep neural networks for secondary structure prediction but is the most accurate according to reported accuracy [98, 99].”
- Transfer String Kernel for Cross-Context DNA-Protein Binding Prediction (2016, background): “Popular bioinformatics tasks predicting proteins’ like local functional properties [20] have been modeled as a labeling or tagging of amino acids on proteins.”