QTLNetMiner - Linking Crop Traits to Genes through Data Integration and Text Mining

Correct identification of causative genes for agronomic traits is valuable for effective marker assisted breeding and reverse genetics. The systematic evaluation of potential functional candidates is, however, time-consuming and requires the integration of many different types of information. These include a range of ontologies for phenotypes, gene function annotations and biochemical pathways and also gene expression data and comparative information from related organisms. An important source of information is gene and phenotype data mined from the scientific literature. We have developed data integration workflows for building genome-scale knowledge networks using the free and open-source Ondex data integration framework (www.ondex.org). QTLNetMiner is a modern web application to interrogate such large genome-scale knowledge networks with trait-based search terms (e.g. early flowering, disease resistance), QTL and gene lists. It allows users to identify and rank potential candidate genes by combining their own data with information from public databases. The user is guided and supported when writing the search terms through features such as real-time user feedback and ontology-based query term suggestions. The relevance of a gene to a particular search query is evaluated using information retrieval methods that take into account the amount of supporting evidence and the specificity of evidence to this gene. The supporting gene-evidence networks are visualized in the KNETViewer which is optimised for visualising and exploring data rich knowledge networks. QTLNetMiner instances for several plant and animal species are available here: https://ondex.rothamsted.ac.uk/QTLNetMiner/.