MAPRes: An efficient method to analyze protein sequence around post‐translational modification sites

Abstract

Functional switches are often regulated by dynamic protein modifications. Assessing protein functions, in vivo, and their functional switches remains still a great challenge in this age of development. An alternative methodology based on in silico procedures may facilitate assessing the multifunctionality of proteins and, in addition, allow predicting functions of those proteins that exhibit their functionality through transitory modifications. Extensive research is ongoing to predict the sequence of protein modification sites and analyze their dynamic nature. This study reports the analysis performed on phosphorylation, Phospho.ELM (version 3.0) and glycosylation, OGlycBase (version 6.0) data for mining association patterns utilizing a newly developed algorithm, MAPRes. This method, MAPRes (Mining Association Patterns among preferred amino acid residues in the vicinity of amino acids targeted for post-translational modifications), is based on mining association among significantly preferred amino acids of neighboring sequence environment and modification sites themselves. Association patterns arrived at by association pattern/rule mining were in significant conformity with the results of different approaches. However, attempts to analyze substrate sequence environment of phosphorylation sites catalyzed for Tyr kinases and the sequence data for O-GlcNAc modification were not successful, due to the limited data available. Using the MAPRes algorithm for developing an association among PTM site with its vicinal amino acids is a valid method with many potential uses: this is indeed the first method ever to apply the association pattern mining technique to protein post-translational modification data.

Phosphoproteome sequence analysis and significance: Mining association patterns around phosphorylation sites utilizing MAPRes

“…Consequently, these general and specific patterns mined at different support levels show the involvement of more than one kinase or its isoforms for catalyzing the phosphorylation of S/T/Y with similar sequence in the vicinity. A comparison of the results of association patterns, mined by MAPRes for the present version (7.0) of Phospho.ELM with those of the previous (i.e., 3.0) reported earlier [Ahmad et al, 2008b] shows that most of the patterns are identical to the previous ones. In case of the patterns mined in the vicinity of phosphorylated S, all previous (28 patterns mined for version 3.0) patterns were included in the present analysis with an addition of five new patterns.…”

“…In this study, phosphorylation data of Phospho.ELM 7.0 [Diella et al, 2008] have been analyzed utilizing MAPRes [Ahmad et al, 2008a] both for all phosphorylated S/T/Y and for different kinase substrate sites. The association patterns mined by MAPRes for all phosphorylated S/T/Y (without kinase information) are consistent with the previous findings with additional patterns compared to the previous analysis [Ahmad et al, 2008b]. Similarly, the patterns mined for kinasespecific phosphorylation sites are also consistent with the previous findings [Obenauer et al, 2003;Blom et al, 2004;Kim et al, 2004;Wang et al, 2008] and also provides novel patterns, to ultimately define the consensus or preferred sequence patterns of substrate required by the different kinases.…”

“…Previously, Phospho.ELM 3.0 [Diella et al, 2004] was analyzed with MAPRes [Ahmad et al, 2008a]. In a previous version of Phospho.ELM 3.0, the kinase data was too scanty to run an analysis for mining association patterns [Ahmad et al, 2008b]. The analysis was therefore restricted to mining association patterns for all phosphorylated S/T/Y without kinase information.…”

“…The strategy adopted for validating the sequence patterns mined by MAPRes, for general and kinase-specific phosphorylation datasets was identical to that performed previously [Ahmad et al, 2008b] for general phosphorylation data analysis of Phospho.ELM 3.0. For this purpose, sequences of 50 proteins, belonging to diverse categories, were downloaded randomly from the SwissProt database [Boeckmann et al, 2003].…”

“…These patterns will be utilized in future for the development and application of an algorithm for training neural networks to achieve maximum prediction accuracy. Previously, an analysis of S/T/Y phosphorylation sites of Phospho.ELM 3.0 [Diella et al, 2004] without kinase information was performed utilizing MAPRes [Ahmad et al, 2008b]. The later version of this database Phospho.ELM 7.0 [Diella et al, 2008] is a phosphorylation sites database that contains 3-4 times more data of experimentally determined protein phosphorylation sites than the previous version (3.0) [Diella et al, 2004].…”