Integration of Bioimage and *Omics data resources
Abstract
This Poster was presented at the 2025 All Hands Meeting of the NFDI4BIOIMAGE Consortium. It presents the current state of data integration activities at the MPI for Evolutionary Biology. Various data and metadata resources such as the internal image data repository OMERO and the Electronic Lab Notebook System OpenBIS are converted into a RDF Knowledge Graph utilizing a R2RML mapping scheme based on the Ontop-VKG framework. The materialized Knowledge Graph is then served via the QLever SPARQL endpoint and user interface. A graphical query editor (SPARNatural) assists users with no SPARQL knowledge in constructing their queries by selecting triple elements from dropdown menus and other widgets. We also present a benchmark comparison of query response times on 10 selected SPARQL queries run against three different endpoint/triplestore implementations. This work is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under the National Research Data Infrastructure – NFDI 46/1 – 501864659
Full text
Integration of Bioimage and *Omics Data Resources Mariana Meireles, Carsten Fortmann-Grote Max Planck Institute for Evolutionary Biology Dept. Microbial Population Biology Integration of Bioimage and *Omics Data Resources Mariana Meireles, Carsten Fortmann-Grote Max Planck Institute for Evolutionary Biology Dept. Microbial Population Biology Annotated Image Data in OMERO Virtualization[3] OME RDF Graph ome_ns:MPB ome_ns:Organism ome_ns:Study ome_core:image rdf:type rdfs:label ome_ns:Strain ome_ns:Investigation ome_core:image ome_core:image ome_core:image rdfs:label ome_core:dataset rdf:type rdfs:label ome_core:dataset ome_core:dataset ome_core:dataset rdf:type ome_core:dataset site_image:41697 "19149" "Pseudomonas fluorescens" "cell measurements" site_dataset:989 site_image:42047 ome_core:Image "MPB19149_OD0....tif" site_project:304 "TimeSeries" "SBW25" site_image:42046 site_dataset:852 site_dataset:806 site_image:42048 "cell shape" site_dataset:789 "MPB19149 15.12.20" ome_core:Project site_dataset:790 ome_core:Dataset Knowledge Graph Construction Scheme USER Query R D F Materialized Knowledge Graph R2RML Virtual Knowledge Graph Endpoint ontology mappings Relational DB SQL to RDF Cache, fast Live query, slow SPARQL to SQL Example query across graph Number of images per samples, QLever UI[2] Sample Collection in OpenBIS Virtualization[3] OpenBIS RDF Graph bisp:375 bisp:376 bisp:391 bisp:393 bisp:374 bisp:400 rdf:type bisp:2 bisp:392 bisp:389 bisp:390 bisp:388 bisp:373 <https://openbis/rdf/Sample/32357> "cell shape defect" "copy of MPB13859 (overnight culture in LB)" "MPB #195" "MPB19149" "SBW25 pbp1a(G1450A)" "PFLU pbp1a(G1450A)" bisc:77 "pbp1a(G1450A)" "A1" "44" "5" "MPB17743" "SBW25" rdfs:label rdfs:label rdfs:label rdfs:label rdfs:label rdfs:label rdfs:label rdfs:label rdfs:label rdfs:label rdfs:label rdfs:label rdfs:label rdfs:label rdfs:label bisp:373 "Strain" bisp:377 "Resistance marker" bisp:402 "Date Created" bisp:372 "sID" bisp:391 "Box" bisp:392 "Position" bisp:375 "Phenotype" bisp:388 "Parent" bisp:390 "Freezer" bisp:378 "Accession" bisp:376 "Construction" bisp:389 "Rack/Drawer" bisp:393 "ID" bisp:374 "Genotype" bisp:2 "Name" Benchmark Queries[1] Basic Retrieval Simple queries fetching all triples or distinct attributes. Useful to establish a baseline response time across the different endpoints. Aggregation Measures efficiency of GROUP BY, ORDER BY, and conditional logic. Filtering Measures efficiency of FILTER and regex for selective retrieval. Complex Multi-Condition Queries Combines BIND, FILTER, and nested conditions. Helpful to identify bottlenecks in query execution under different architectures. Full Dataset Retrieval Stress tests performance with large-scale retrieval throughout the entire dataset. 1.50 1.75 2.00 2.25 2.50 2.75 3.00 time (s) query = list of attributes 2 3 4 5 6 7 query = group by attribute sort 2 3 4 5 6 7 8 9query = group by attribute avg 2.2 2.4 2.6 2.8 3.0 query = group by attribute other avg 20 25 30 35 40 45 50 query = image properties ontop fuseki virtuoso endpoint 2 3 4 5 6 7 time (s) query = count triples ontop fuseki virtuoso endpoint 10 20 30 40 50 query = construct triples ontop fuseki virtuoso endpoint 2 3 4 5 6 7 query = fulltext search ontop fuseki virtuoso endpoint 2.2 2.4 2.6 2.8 3.0 3.2 3.4 query = regex ontop fuseki virtuoso endpoint 1.6 1.8 2.0 2.2 2.4 2.6 query = filter regex sort 1 k 10 k 100 k ntriples 0 20 40 60 time (s) clock = Wall (s) 1 k 10 k 100 k ntriples 2 4 6 8 clock = User (s) 1 k 10 k 100 k ntriples 0.25 0.50 0.75 1.00 1.25 1.50 clock = Sys (s) query list of attributes group by attribute sort group by attribute avg group by attribute other avg image properties count triples construct triples fulltext search regex filter regex sort Graphical query editor SPARNatural[4] References [1] Angus Addlesee. Comparing Linked Data Triplestores. en. Jan. 2019. url:https://medium.com/ wallscope/comparing-linked-data-triplestores-ebfac8c3ad4f (visited on 01/31/2025). [2] Hannah Bast and Bj¨ orn Buchhold. QLever. A Query Engine for Efficient SPARQL+Text Search. 2017. doi:10.1145/3132847.3132921. [3] Diego Calvanese et al. “Ontop: Answering SPARQL Queries over Relational Databases”. In: Semantic Web Journal (2015), 1278–2490. url:http://www.semantic-web-journal.net/content/ontopanswering-sparql-queries-over-relational-databases-1. [4] Thomas Francart.“Sparnatural: A Visual Knowledge Graph Exploration Tool”. In: The Semantic Web: ESWC 2023 Satellite Events. Springer Nature Switzerland, 2023, pp. 11–15. isbn: 9783031434587. doi: 10.1007/978-3-031-43458-7_2. This work is licensed under CC-BY-SA 4.0 PDF copy: