Recognition of sites of functional specialisation in all known eukaryotic protein kinase families

PLoS Comput Biol. 2018 Feb 13;14(2):e1005975. doi: 10.1371/journal.pcbi.1005975. eCollection 2018 Feb.


The conserved function of protein phosphorylation, catalysed by members of protein kinase superfamily, is regulated in different ways in different kinase families. Further, differences in activating triggers, cellular localisation, domain architecture and substrate specificity between kinase families are also well known. While the transfer of γ-phosphate from ATP to the hydroxyl group of Ser/Thr/Tyr is mediated by a conserved Asp, the characteristic functional and regulatory sites are specialized at the level of families or sub-families. Such family-specific sites of functional specialization are unknown for most families of kinases. In this work, we systematically identify the family-specific residue features by comparing the extent of conservation of physicochemical properties, Shannon entropy and statistical probability of residue distributions between families of kinases. An integrated discriminatory score, which combines these three features, is developed to demarcate the functionally specialized sites in a kinase family from other sites. We achieved an area under ROC curve of 0.992 for the discrimination of kinase families. Our approach was extensively tested on well-studied families CDK and MAPK, wherein specific protein interaction sites and substrate recognition sites were successfully detected (p-value < 0.05). We also find that the known family-specific oncogenic driver mutation sites were scored high by our method. The method was applied to all known kinases encompassing 107 families from diverse eukaryotic organisms leading to a comprehensive list of family-specific functional sites. Apart from other uses, our method facilitates identification of specific protein interaction sites and drug target sites in a kinase family.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adenosine Triphosphate / chemistry
  • Animals
  • Binding Sites*
  • Catalytic Domain
  • Cluster Analysis
  • Computational Biology
  • Cyclin-Dependent Kinases / metabolism*
  • Entropy
  • Humans
  • MAP Kinase Signaling System*
  • Models, Statistical
  • Mutation
  • Phosphorylation
  • Phylogeny
  • Probability
  • Protein Binding
  • Protein Domains
  • Protein Kinases / metabolism*
  • Protein Serine-Threonine Kinases / metabolism*
  • ROC Curve
  • Substrate Specificity


  • Adenosine Triphosphate
  • Protein Kinases
  • Protein Serine-Threonine Kinases
  • Cyclin-Dependent Kinases

Grants and funding

This research was supported by Indian Institute of Science-Department of Biotechnology partnership program as well as by the Mathematical Biology project sponsored by Department of Science and Technology (DST) and Indo-French Centre for the Promotion of Advanced Research (IFCPAR / CEFIPRA) grant (5203-2). Support for Infrastructural facilities from Fund for Improvement of Science and Technology infrastructure (FIST), DST, Ministry of Human Resource Development (MHRD) and Centre for Advanced Study (CAS), University Grants Commission (UGC) is also acknowledged. NS is a J C Bose National Fellow supported by DST. RK was supported by Indo-French Centre for the Promotion of Advanced Research (IFCPAR / CEFIPRA; no. 5203-2). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.