Mining Associations for Organism Characteristics in Prokaryotes - an Integrative Approach

Correlations and associations between specific organism characteristics (such as genome size, genome GC content, optimal growth temperature, habitat, oxygen requirements) may provide for deeper comprehension of evolutionary processes as well as for some prediction possibilities, e.g., trends prediction of some pandemia. There is a plenty of genotype data and gene sequences for different organisms, which is usually well structured and deposited into databases. On the other side, data on phenotypic characteristics of organisms are often scattered across different text documents, e.g., scientific papers or encyclopedias. We reconsider correlations between organism characteristics for superkingdoms Archaea and Bacteria and extend the study in a number of ways. We use a larger dataset of prokaryotes as well as a larger set of characteristics by integrating several existing databases with data obtained by literature mining. We recalculate some highexpectation correlations between genomic characteristics (genome size, GC content, distribution among functional groups of proteins) and apply algorithms for association rule mining in order to identify the most confident associations between specific modalities of both genotype and phenotype characteristics.