In life sciences, there has been a long-standing effort of standardization and integration of reference datasets and databases. Despite these efforts, many studies data are provided using specific and non-standard formats. This hampers the capacity to reuse the studies data in other pipelines, the capacity to reuse the pipelines results in other studies, and the capacity to enrich the data with additional information. The Regulatory Circuits project is one of the largest efforts for integrating human cell genomics data to predict tissue-specific transcription factor-genes interaction networks. In spite of its success, it exhibits the usual shortcomings limiting its update, its reuse (as a whole or partially), and its extension with new data samples. To address these limitations, the resource has previously been integrated in an RDF triplestore so that TF-gene interaction networks could be generated with two SPARQL queries. However, this triplestore did not store the computed networks and did not integrate metadata about tissues and samples, therefore limiting the reuse of this dataset. In particular, it does not enable to reuse only a portion of Regulatory Circuits if a study focuses on a subset of the tissues, nor to combine the samples described in the datasets with samples from other studies. Overall, these limitations advocate for the design of a complete, flexible and reusable representation of the Regulatory Circuits dataset based on Semantic Web technologies.
We provide a modular RDF representation of the Regulatory Circuits, called Linked Extended Regulatory Circuits (LERC). It consists in (i) descriptions of biological and experimental context mapped to the references databases, (ii) annotations about TF-gene interactions at the sample level for 808 samples, (iii) annotations about TF-gene interactions at the tissue level for 394 tissues, (iv) metadata connecting the knowledge graphs cited above. LERC is based on a modular organisation into 1,205 RDF named graphs for representing the biological data, the sample-specific and the tissue-specific networks, and the corresponding metadata. In total it contains 3,910,794,050 triples and is available as a SPARQL endpoint.
The flexible and modular architecture of LERC supports biologically-relevant SPARQL queries. It allows an easy and fast querying of the resources related to the initial Regulatory Circuits datasets and facilitates its reuse in other studies. ASSOCIATED WEBSITE: https://regulatorycircuits-lod.genouest.org....

Improving reusability along the data life cycle: a regulatory circuits case study

10.1186/s13326-022-00266-4