Distinguishing fungal from bacterial infection : a mixed integer linear programming approach / von Joao Pedro Leonor Fernandes Saraiva

The immune system is responsible for protecting the host from infections. In healthy individuals, this system is generally able to fight and clear any pathogen it encounters. Blood stream infections can be caused by several pathogens such as viruses, fungi and bacteria. Delivery of appropriate treatment requires rapid identification of the invading pathogen. The use of in situ experiments attempts to identify pathogen specific immune responses but these often lead to heterogeneous biomarkers due to the high variability in methods and materials used. Support Vector Machines (SVMs) allow using gene expression patterns to discriminate between two types of infection. Comparing gene lists from independent studies shows a high degree of inconsistency. To produce consistent gene signatures, capable of discriminating fungal from bacterial infection, SVMs using Mixed Integer Linear Programming (MILP) were employed allowing the combination of classifiers using different datasets. Employing this method demonstrated the improvement in consistency of the produced gene signatures that distinguished fungal from bacterial infections irrespective of the type of the leukocyte or the experimental setup. The produced biomarker list showed an increase in consistency of 42% when compared to single classifiers and predicted the infecting pathogen on an unseen dataset with an average accuracy of 87%. Restricting the analysis to datasets comprised of peripheral blood mononuclear cells and monocytes, showed an enrichment of genes from the lysosome pathway that was not shown when using independent classifiers. Moreover, the results suggested that the lysosome pathway is specifically induced in monocytes. In conclusion, the combined classifier approach increased the consistency of the gene signatures, compared to single classifiers and "unmasked" the monocyte-specific expression profile for fungal infections.

Saved in:
Person: Saraiva, Joao [Author]
Corporate Author: Friedrich-Schiller-Universität Jena [Degree granting institution]
Format: Book
Language note:Zusammenfassungen in deutscher und englischer Sprache
Publication:Jena, 2018
Printing place:Jena
Dissertation Note:Dissertation, Friedrich-Schiller-Universität Jena, 2018
Subjects:Infektion > Genexpression > Immunozyt > Detektion
Type of content:Hochschulschrift
Related resources:Erscheint auch als Online-Ausgabe: Distinguishing fungal from bacterial infection
Physical description:126 Blätter : Illustrationen, Diagramme ; 29,5 cm
Basic Classification: 44.45 Immunologie
44.75 Infektionskrankheiten, parasitäre Krankheiten
42.13 Molekularbiologie