KnetMiner 5.0: richer, faster, smarter

The new KnetMiner 5.0 comes with enhancements, speed improvements and new free/pro features.

Powerful search queries

The gene list search has been accelerated and enhanced to support wildcards, e.g., users can easily search for all TPS* or PSY* genes to get all members of a big gene family. Furthermore, a match can now be against one of the gene synonyms because crop gene names in KnetMiner are extrapolated via Arabidopsis orthologs and the UniProtKB-Swissprot best matches. The keyword search still supports the full Lucene query syntax. Combining both the gene list and keyword searches together allows users to ask powerful questions: for example, we can easily check which wheat TPS* genes respond to specific biotic or abiotic stresses: “defense response” OR ((heat OR drought) AND tolerance). Additional features in the search result feedback now show both the number of linked and unlinked genes w.r.t the user keywords, and the time taken for the search to finish, e.g. “In total 136 linked genes and 16 unlinked genes were found (0.58 seconds).” As before, searching for “drought tolerance” will match different types of nodes in the knowledge graph such as publications and ontology terms. In KnetMiner 5.0, one can also search for links between genes and a specific trait ontology node. We can use following syntax: genelist:TPS* and keyword: “ConceptID:4346721”.

Gene set enrichment analysis

We have enhanced the Evidence View with a new feature to create a knowledge network for enriched terms. When a user provides a genelist, KnetMiner performs a knowledge graph-based enrichment analysis and presents the over-represented terms with a p-value. KnetMiner 5.0 allows users to select multiple terms of interest in Evidence View and render the combined linkages and evidences in Network View. The benefits are twofold, we can extend a classical enrichment analysis to any knowledge type traits, GO terms, pathways, protein domains, transcription factors etc., and also present the full provenance to the user. Also, the full Evidence View table can now be downloaded from KnetMiner and iterative searches are now more seamless with new features to add term ids to the search box.

Improved network visualisation

The network visualisation (KnetMaps) now uses advanced node labelling to highlight important parts of the sub-graph. This helps users distinguish gene nodes from different species, which is highly useful for multi-species KGs. “Dashed” edges are used to flag relationships that are not absolute and need to be treated with caution. Networks that would have been too large are also auto-filtered before rendering. A new “export” feature allows users to download all visible genes from the network in tabular format.

New and updated datasets

The underlying knowledge graph datasets have been updated to Ensembl Genomes Plants 51 releases and latest data snapshots from UniPort, PubMed, BioGrid and other data sources. Public KnetMiner resources (e.g. wheat) integrate with a Arabidopsis knowledge graph via Ensembl Compara homologies. Even richer multi-species knowledge graph bundles and KnetMiner resources (e.g. poaceae, solanaceae) are also now available for a subscription fee. Our public RDF-SPARQL graph datasets and endpoints have also been updated and improved. The latest KnetMiner publication and COVID-19 KnetMiner data note are out now as well.

Free and Pro features

Public KnetMiner resources such as wheat, Arabidopsis and COVID-19 are offered in two flavours: “Free” and “Pro”. The free version is the perfect starting point for anyone interrogating the platform with only a small number of genes. The free plan allows users to query KnetMiner with max. 20 genes of interest and then visualize a network of up to 10 genes. Upgrading to the Pro version enables users to be more efficient in their analysis. Premium users will be able to search with unlimited genes, create large networks for up to 200 genes and store unlimited gene networks.

Further improvements

The KnetMiner “genepage API now also analyses the user query in real-time. It uses keywords and a gene list provided by the user to generate a visual word summary of what the resulting network contains. This provides an intuitive instant overview of the nodes and relationships within the network.

KnetMiner 5.0 also provides enhanced configuration options for on-premise deployment. Admins can enable/disable or re-route google analytics statistics and save anonymized queries to an Amazon S3 bucket for internal dashboards and usage metrics. The wiki has been updated with more information for developers and users.

For further information and detailed release notes, visit our GitHub pages.

We hope you enjoy the new KnetMiner 5.0 experience.