PGxO and PGxLOD: a reconciliation of pharmacogenomic knowledge of various provenances, enabling further comparison

Abstract

Background Pharmacogenomics (PGx) studies how genomic variations impact variations in drug response phenotypes. Knowledge in pharmacogenomics is typically composed of units that have the form of ternary relationships gene variant -drug -adverse event. Such a relationship states that an adverse event may occur for patients having the specified gene variant and being exposed to the specified drug. State-of-the-art knowledge in PGx is mainly available in reference databases such as PharmGKB and reported in scientific biomedical literature. But, PGx knowledge can also be discovered from clinical data, such as Electronic Health Records (EHRs), and in this case, may either correspond to new knowledge or confirm state-of-the-art knowledge that lacks "clinical counterpart" or validation. For this reason, there is a need for automatic comparison of knowledge units from distinct sources. Results In this article, we propose an approach, based on Semantic Web technologies, to represent and compare PGx knowledge units. To this end, we developed PGxO, a simple ontology that represents PGx knowledge units and their components. Combined with PROV-O, an ontology developed by the W3C to represent provenance information, PGxO enables encoding and associating provenance information to PGx relationships. Additionally, we introduce a set of rules to reconcile PGx knowledge, i.e. to identify when two relationships, potentially expressed using different vocabularies and levels of granularity, refer to the same, or to different knowledge units. We evaluated our ontology and rules by populating PGxO with knowledge units extracted from PharmGKB (2,701), the literature (65,720) and from discoveries reported in EHR analysis studies (only 10, manually extracted); and by testing their similarity. We called PGxLOD (PGx Linked Open Data) the resulting knowledge base that represents and reconciles knowledge units of those various origins. Conclusions The proposed ontology and reconciliation rules constitute a first step toward a more complete framework for knowledge comparison in PGx. In this direction, the experimental instantiation of PGxO, named PGxLOD, illustrates the ability and difficulties of reconciling various existing knowledge sources. BackgroundIn this article, we present a simple ontology named PGxO, developed to reconcile and trace knowledge in pharmacogenomics (PGx). We instantiated this ontology with knowledge of various origins to both 1/23 illustrate the relevance of the ontology and constitute a Linked Open Data (LOD) data set for PGx [5].PGx itself studies how genomics impact individual variations in drug response phenotypes [63]. Knowledge in pharmacogenomics is of particular interest for the implementation of personalized medicine, i.e. a medicine tailoring treatments (chosen drugs and dosages) to every patient, in order to reduce the risk of adverse events. Indeed, best known examples of PGx knowledge already led to the development of clinical guidelines and practices [9] that recommend considering individual genotype when ...

Discovering alignment relations with Graph Convolutional Networks: A biomedical case study

“…We experimented with PGxLOD 2 , a large knowledge graph about pharmacogenomics (PGx) that we previously built [16]. Our approach is implemented in Python, using PyTorch and the Deep Graph Library for learning embeddings, and scikit-learn for clustering.…”

“…We detail the core of our matching approach (node embeddings and clustering) in Section 3, and how inference rules associated with domain knowledge are considered in Section 4. In Section 5, we experiment this approach on PGxLOD, a large knowledge graph we built that contains 50,435 PGx relationships [16]. Finally, we discuss our results and conclude in Section 6 and 7.…”

Tackling scalability issues in mining path patterns from knowledge graphs: a preliminary study

“…We explore PGxLOD 14 [10], a knowledge graph that aggregates several sets of Linked Open Data (LOD) describing drugs, phenotypes, and genetic factors: PharmGKB, ClinVar, DrugBank, SIDER, DisGeNET, and CTD. This aggregation may lead to features combining units from several LOD sets.…”

“…P h is explained in Subsection 3.3.4 10. Additionally, to avoid loops, P can only be expanded at iteration h with individuals ve such that there exists at least one seed vertex in SUPPORTSET(P ) whose shortest distance to ve is h.11 Not all generated path patterns remain in the dependency structure at the end of an iteration, see Subsections 3.3.3 and 3.3.4.…”

Experiments on transfer learning architectures for biomedical relation extraction

“…Indeed, RE allows to extract and structure elements of knowledge from natural language texts by identifying and typing the relationships that may be mentioned between named entities (Pawar et al, 2017). The resulting relations can be normalized and assembled in the form of a knowledge graph (KG), which summarizes a domain of knowledge and may serve as a structured intermediate for subsequent tasks of knowledge discovery or knowledge comparison (Monnin et al, 2019). In this work, we explore the task of RE, in the biomedical domain, and from a deep learning perspective.…”