@ARTICLE{10.3389/fmicb.2015.00764, AUTHOR={Remmele, Christian W. and Luther, Christian H. and Balkenhol, Johannes and Dandekar, Thomas and Müller, Tobias and Dittrich, Marcus T.}, TITLE={Integrated inference and evaluation of host–fungi interaction networks}, JOURNAL={Frontiers in Microbiology}, VOLUME={6}, YEAR={2015}, URL={https://www.frontiersin.org/articles/10.3389/fmicb.2015.00764}, DOI={10.3389/fmicb.2015.00764}, ISSN={1664-302X}, ABSTRACT={Fungal microorganisms frequently lead to life-threatening infections. Within this group of pathogens, the commensal Candida albicans and the filamentous fungus Aspergillus fumigatus are by far the most important causes of invasive mycoses in Europe. A key capability for host invasion and immune response evasion are specific molecular interactions between the fungal pathogen and its human host. Experimentally validated knowledge about these crucial interactions is rare in literature and even specialized host–pathogen databases mainly focus on bacterial and viral interactions whereas information on fungi is still sparse. To establish large-scale host–fungi interaction networks on a systems biology scale, we develop an extended inference approach based on protein orthology and data on gene functions. Using human and yeast intraspecies networks as template, we derive a large network of pathogen–host interactions (PHI). Rigorous filtering and refinement steps based on cellular localization and pathogenicity information of predicted interactors yield a primary scaffold of fungi–human and fungi–mouse interaction networks. Specific enrichment of known pathogenicity-relevant genes indicates the biological relevance of the predicted PHI. A detailed inspection of functionally relevant subnetworks reveals novel host–fungal interaction candidates such as the Candida virulence factor PLB1 and the anti-fungal host protein APP. Our results demonstrate the applicability of interolog-based prediction methods for host–fungi interactions and underline the importance of filtering and refinement steps to attain biologically more relevant interactions. This integrated network framework can serve as a basis for future analyses of high-throughput host–fungi transcriptome and proteome data.} }