Protein interaction sentence detection using multiple semantic kernels

Abstract

BackgroundDetection of sentences that describe protein-protein interactions (PPIs) in biomedical publications is a challenging and unresolved pattern recognition problem. Many state-of-the-art approaches for this task employ kernel classification methods, in particular support vector machines (SVMs). In this work we propose a novel data integration approach that utilises semantic kernels and a kernel classification method that is a probabilistic analogue to SVMs. Semantic kernels are created from statistical information gathered from large amounts of unlabelled text using lexical semantic models. Several semantic kernels are then fused into an overall composite classification space. In this initial study, we use simple features in order to examine whether the use of combinations of kernels constructed using word-based semantic models can improve PPI sentence detection.ResultsWe show that combinations of semantic kernels lead to statistically significant improvements in recognition rates and receiver operating characteristic (ROC) scores over the plain Gaussian kernel, when applied to a well-known labelled collection of abstracts. The proposed kernel composition method also allows us to automatically infer the most discriminative kernels.ConclusionsThe results from this paper indicate that using semantic information from unlabelled text, and combinations of such information, can be valuable for classification of short texts such as PPI sentences. This study, however, is only a first step in evaluation of semantic kernels and probabilistic multiple kernel learning in the context of PPI detection. The method described herein is modular, and can be applied with a variety of feature types, kernels, and semantic models, in order to facilitate full extraction of interacting proteins.

Popular Computational Methods to Assess Multiprotein Complexes Derived From Label-Free Affinity Purification and Mass Spectrometry (AP-MS) Experiments

“…HPRD (73), IntAct, (74) MINT (75), DIP (76), BIND (77), BioGRID (78), MPact (79), and InnateDB (80)), collect interaction data via direct submission, expert curation, or computational text-mining of protein interaction data from publications under curator supervision. In addition to the limitations associated with automated text-mining (81,82), one report suggests that any two databases fully agree on only 42% of interaction data and 62% of proteins curated from the same publication (83). It is demonstrated, however, that such discrepancies are due predominantly to different curation methods, e.g.…”

When One and One Gives More than Two: Challenges and Opportunities of Integrative Omics

“…One way to address this issue is the development of so-called “data warehouses,” in which a significant effort is being put in by developers a priori to store and integrate heterogeneous primary databases into a coherent scheme by making use of intermediate abstraction layers between the raw data layer and the user access layer (Rhodes et al, 2004; Chen et al, 2010). An alternative promising approach to data integration in life sciences is offered by Semantic Web technologies (Splendiani et al, 2011). These technologies enable an immediate “connection” between data, which can be easily queried across different databases.…”

An empirical analysis of ontology reuse in BioPortal

“…Mortensen et al [1] investigated the use of ontology design patterns among the BioPortal ontologies, including the use of upper-level ontologies such as the BFO. Ghazvinian et al [11] looked at the orthogonality of the OBO Library [5] ontologies and their extent of reuse. Kamdar et al [2] reviewed term (i.e., class) reuse among the BioPortal ontologies.…”