MusiteDeep: a deep-learning framework for general and kinase-specific phosphorylation site prediction

Wang, Duolin; Zeng, Shuai; Xu, Chunhui; Qiu, Wang-Ren; Joshi, Trupti; Xu, Dong

doi:10.1093/bioinformatics/btx496

Cited by 220 publications

(221 citation statements)

References 30 publications

Supporting

Mentioning

207

Contrasting

Order By: Relevance

“…As underlined by a recent review [7], most of the phosphoproteome is uncharted: more than 95% of reported human phosphosites have no known kinase or associated biological function. 45 To identify a phosphorylation site on a protein sequence, several computational methods have been proposed earlier [13,15,[17][18][19][20][24][25][26][27][28][29][30][31][32]. Since these methods can also provide kinase specific predictions, they can be used to predict associated kinases of a known phosphosite.…”

Section: Introductionmentioning

confidence: 99%

“…The application of such tools is limited to kinases for which a substantial number of target phosphosites are available for training. For 55 example, MusiteDeep [20], uses deep learning to predict binding sites for kinases, and it exclusively focuses on kinase families with at least 100 experimentally verified phosphosites. Recently, the use of phosphorylation data to predict kinases has been proposed, but these methods also require knowledge of target sites for a kinase to make predictions for that kinase [33].…”

Section: Introductionmentioning

confidence: 99%

“…35 Advances in mass spectrometry-based phosphoproteomics has enabled identification of phosphosites at the proteome level [10][11][12]. Many computational models have also been developed to predict phosphosites in a given input protein sequence (recently [13][14][15][16][17][18][19][20] and earlier methods reviewed in [21]). Once a phosphosite is identified, either experimentally or computationally, determining the kinase that is responsible for catalyzing the phosphorylation of this site becomes the key question.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

DeepKinZero: Zero-Shot Learning for Predicting Kinase-Phosphosite Associations Involving Understudied Kinases

Deznabi

Arabaci

Koyutürk

et al. 2019

Preprint

View full text Add to dashboard Cite

5Protein phosphorylation is a key regulator of protein function in signal transduction pathways. Kinases are the enzymes that catalyze the phosphorylation of other proteins in a target specific manner. The dysregulation of phosphorylation is associated with many diseases including cancer. Although the advances in phosphoproteomics enable the identification of phosphosites at the proteome level, most of the phosphoproteome is still in the dark: more than 95% of 10 reported human phosphosites have no known kinases. Determining which kinase is responsible for phosphorylating a site remains an experimental challenge. Existing computational methods require several examples of known targets of a kinase to make accurate kinase specific predictions, yet for a large body of kinases, only a few or no target sites are reported. We present DeepKinZero, the first zero-shot learning approach to predict the kinase acting on a phosphosite for kinases with 15 no known phosphosite information. DeepKinZero transfers knowledge from kinases with many known target phosphosites to those kinases with no known sites through a zero-shot learning model. The kinase specific positional amino acid preferences are learned using a bidirectional recurrent network. We show that DeepKinZero achieves significant improvement in accuracy for kinases with no known phosphosites in comparison to the baseline model. By expanding our 20 knowledge on understudied kinases, DeepKinZero can help to chart the phosphoproteome atlas.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

DeepKinZero: Zero-Shot Learning for Predicting Kinase-Phosphosite Associations Involving Understudied Kinases

Deznabi

Arabaci

Koyutürk

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…That is, for a given kinase family, a separate machine learning model was built to capture the unique pattern embedded in the substrate sequences of this family. Many successful applications belong to this family-specific category, such as NetPhosK (Neural Networks) [4] , KinasePhos (Hidden Markov Models) [5], PhosPhoPICK (Bayesian Networks) [6], PhosphoPredict (Random Forests) [7], Musit-eDeep (Convolutional Neural Network) [8], and many others [9][10][11][12][13]. An important prerequisite of these family-specific methods is the availability of a sufficient amount of high-quality data for model training.…”

Section: Introductionmentioning

confidence: 99%

“…However, this is often not the case: a majority of kinase families have less than 30 experimentally verified phosphorylation sites. As a result, previous studies have to choose to build family-specific models for well-characterized kinase families (typically <10 families) with abundant training data [7,8,14]. For those less-studied kinase families, using the family-specific approach fails to provide satisfactory performance due to the lack of sufficient training data.…”

Section: Introductionmentioning

confidence: 99%

Mitigating Data Scarcity in Protein Binding Prediction Using Meta-Learning

Luo

Zhao

et al. 2019

Preprint

View full text Add to dashboard Cite

A plethora of biological functions are performed through various types of protein-peptide binding. Prime examples include the protein kinase phosphorylation on peptide substrates and the binding of major histocompatibility complex to neoantigens in the immune system. Understanding the specificity of protein-peptide interactions is critical for unraveling the architectures of functional pathways and the mechanisms of cellular processes in human cells. Despite massspectrometric techniques were developed for the identification of protein-peptide interactions, our understanding of the preferences of proteins on their binding peptides is still rudimentary. As a complementary direction, a line of computational prediction methods has been recently proposed to predict protein-peptide bindings which efficiently provide rich functional annotations on a large scale. To achieve a high prediction accuracy, these computational methods require a sufficient amount of data to build the prediction model. However, the number of experimentally verified protein-peptide bindings is often limited in real cases. For example, a majority of protein kinases have very few experimentally verified phosphorylation sites (e.g., less than 30 sites) in existing databases. These methods are thus limited to building accurate prediction models for only well-characterized proteins with a large volume of known binding peptides and cannot be extended to predict new binding peptides for less-studied proteins. In this paper, we introduce a generic framework to address this issue of data scarcity in protein binding prediction. We demonstrate the applicability of our framework in predicting kinase-specific phosphorylation sites. Our method uses an effective training strategy to build a prediction model with robust transferability. The model is able to predict the phosphorylation sites of a less-studied kinase, even if there is only a small number of phosphorylation sites known for this kinase. To achieve this, we train the model via a meta-learning phase followed by a few-shot learning phase. We demonstrate our framework has better transferability than state-of-the-art methods and is effective in utilizing limited data to accurately predict phosphorylation sites for less-characterized kinases. The implementation of our framework is available at https://github.com/luoyunan/MetaKinase.

show abstract

Importance of protein Ser/Thr/Tyr phosphorylation for bacterial pathogenesis

et al. 2020

View full text Add to dashboard Cite

Protein phosphorylation regulates a large variety of biological processes in all living cells. In pathogenic bacteria, the study of serine, threonine, and tyrosine (Ser/Thr/Tyr) phosphorylation has shed light on the course of infectious diseases, from adherence to host cells to pathogen virulence, replication, and persistence. Mass spectrometry (MS)-based phosphoproteomics has provided global maps of Ser/Thr/Tyr phosphosites in bacterial pathogens. Despite recent developments, a quantitative and dynamic view of phosphorylation events that occur during bacterial pathogenesis is currently lacking. Temporal, spatial, and subpopulation resolution of phosphorylation data is required to identify key regulatory nodes underlying bacterial pathogenesis. Herein, we discuss how technological improvements in sample handling, MS instrumentation, data processing, and machine learning should improve bacterial phosphoproteomic datasets and the information extracted from them. Such information is expected to significantly extend the current knowledge of Ser/ Thr/Tyr phosphorylation in pathogenic bacteria and should ultimately contribute to the design of novel strategies to combat bacterial infections.

show abstract

MusiteDeep: a deep-learning framework for general and kinase-specific phosphorylation site prediction

Abstract: Supplementary data are available at Bioinformatics online.

Cited by 220 publications

References 30 publications

DeepKinZero: Zero-Shot Learning for Predicting Kinase-Phosphosite Associations Involving Understudied Kinases

DeepKinZero: Zero-Shot Learning for Predicting Kinase-Phosphosite Associations Involving Understudied Kinases

Mitigating Data Scarcity in Protein Binding Prediction Using Meta-Learning

Importance of protein Ser/Thr/Tyr phosphorylation for bacterial pathogenesis

Contact Info

Product

Resources

About