Text Mining of Protein Phosphorylation Information Using a Generalizable Rule-Based Approach

Abstract

Literature-based annotation of protein phosphorylation is the focus of many biological databases, as phosphorylation is a global regulator of cellular activity. To speed up manual curation of phosphorylation information, text mining technology has been utilized. In this paper, we report our ongoing effort to enhance RLIMS-P, a rule-based information extraction (IE) system to identify protein phosphorylation information in scientific literature. Despite the high accuracy attained by RLIMS-P, the use of elaborated patterns and rules resulted in a substantial effort for system development and maintenance. To mitigate this challenge, we redesigned RLIMS-P and integrated new natural language processing (NLP) techniques. It has also been adapted to mine full-text articles and generalized to be able to exploit common features for different post-translational modifications (PTMs). The updated RLIMS-P (version 2.0) was evaluated on abstracts in the publicly available BioNLP GENIA event extraction (GE) corpus, and achieved F-scores of 0.92 and 0.96 for phosphorylation substrate and site, respectively. On a full-text corpus developed in-house, it achieved F-scores of 0.91 and 0.92 for substrate and site, and 0.88 for kinase. The system was applied to the PubMed Central (PMC) Open Access Subset, and promising results have been obtained in mining the full-text articles. RLIMS-P focuses on protein phosphorylation information, but its new design would be generalizable for other PTM types. RLIMS-P version 2.0 is available at: http://proteininformationresource.org/rlimsp/.

RLIMS-P 2.0: A Generalizable Rule-Based Information Extraction System for Literature Mining of Protein Phosphorylation Information

“…As seen in the study by Landeghem et al [27], the selection of the corpus could have a significant impact on the performance of biological event extraction systems, including performance on phosphorylation events. In our prior work [11], we evaluated RLIMS-P 2.0 on the 2011 GE corpus. We further examined the evaluation results and analyzed the patterns pertaining to phosphorylation events in the 2011 GE corpus as detailed in Section 4.2.4.…”

“…Apart from the low-level text processing implemented, we facilitate mining of phosphorylation information by treating each article section as a separate document, just as the MEDLINE abstract. Writing styles in the full-text articles, however, can be different from those in abstracts [43] and that might affect the performance of RLIMS-P. To evaluate the system in processing full-text articles, we prepared an annotated full-text corpus for our earlier work [11]. We report the development of this corpus, and include the RLIMS-P 2.0 evaluation results on this corpus (Table 2).…”

“…In our previous study [11], we evaluated RLIMS-P 2.0 on the MEDLINE abstracts of the GENIA event extraction (GE) corpus, which was released for the 2011 BioNLP shared task (BioNLP-ST). In this corpus, however, expressions used to report phosphorylation events were limited, and hence insufficient for developing and thoroughly evaluating phosphorylation IE systems.…”

Construction of phosphorylation interaction networks by text mining of full-length articles using the eFIP system

“…( 12 ), and the systems that participated in the BioNLP 2011 Shared Task ( 13 ). We chose the RLIMS-P system ( 6 , 7 ) because it has been evaluated with a corpus covering a wide variety of expressions describing phosphorylation events; it extracts information from multiple sentences; and it has recently been improved with new generalizable rules that boost its performance and allow for the possibility of extending to other post-translational modifications.…”

“…Addressing the users’ feedback, the main contributions of this work are: (i) full-scale processing of full-length articles from the PubMed Central open-access (PMC OA) database; (ii) the enhancement of the PPI module to include additional words/phrases for PPIs; (iii) the inclusion in the pipeline of iSimp, a sentence simplifier ( 4 , 5 ) to improve the recall when extracting phosphorylation–PPI relations; (iv) the incorporation of the latest version of the RLIMS-P system [RLIMS-P 2.0 ( 6 , 7 )] for phosphorylation event extraction; (v) the enhancement of the website, which allows a user to search for specific kinases, substrates, interacting proteins, keywords or lists of document IDs; (vi) the visualization of the network of interacting proteins via the Cytoscape package ( 8 ); (vii) the inclusion of gene normalization via the GenNorm system ( 9 ); (viii) an evaluation of the eFIP system on full-length articles and (ix) a corpus of annotated data from 100 randomly chosen sections from full-length documents, containing 272 unique annotations.…”