GENOME-WIDE PRIORITIZATION OF CANDIDATE DIAGNOSTIC ANTIGENIC MARKERS IN HUMAN PATHOGENS

RAMOA D; BRUNER M; CARMONA SJ; AGÜERO F

Ciudad Autonoma de Buenos Aires

Conferencia; 4th ISCB Latin America Conference / 7th Argentinian Congress of Bioinformatics; 2016

International Society for Computational Biology (ISCB) / Asociacion Argentina de Bioinformatica y Biologia Computacional (A2B2C)

GENOME-WIDE PRIORITIZATION OF CANDIDATE DIAGNOSTIC ANTIGENIC MARKERS IN HUMAN PATHOGENS Diego Ramoa1, Mauricio Brunner1, Santiago J Carmona2,3, Fernán Agüero2 1 Facultad de Ingeniería, Universidad Nacional de Entre Ríos, Oro Verde, Entre Rios, Argentina2 Instituto de Investigaciones Biotecnológicas ? Instituto Tecnológico de Chascomús, Universidad de San Martín ? Consejo Nacional de Investigaciones Científicas y Técnicas, San Martín, B1650HMP, San Martín, Buenos Aires, Argentina3 Present address: Ludwig Center for Cancer Research - University of Lausanne - Swiss Institute of Bioinformatics, Lausanne, Switzerland.Peptide antigens offer a smart solution for the development of next-generation diagnostics. In contrast with protein antigens that need to be obtained as structurally stable and intact proteins, peptide antigens provide other advantages such as: they are well-defined from the start; only information about the protein sequence is needed; they have much lower chances of cross-reactivity; and they are fully synthetic, which is cost-effective with usually very short supply times. Furthermore, today peptide microarrays are a highly efficient platform for discovery and mapping of candidate antigenic peptides. In this work we present a bioinformatics strategy to filter and reduce the number of candidate peptides from complete pathogen proteomes. For this our pipeline decomposes a proteome into peptides of user-defined length, and analyzes 11 features related to their antigenic potential, predicted subcellular location, secondary structures, and sequence similarity with other pathogen species, amongst others. An important novelty of the method is that it analyzes features that are new to epitope prediction such as subcellular localization or sequence similarity to host proteins or to proteins from co-endemic pathogens. To validate our tool, we prioritized candidate antigens for 10 human pathogens (Borrelia burgdoferi, Francisella tularensis, Brucella melitensis, Coxiella burnetii, Leptospira interrogans, Mycobacterium tuberculosis and M. leprae, Plasmodium falciparum, Toxoplasma gondii and Trypanosoma cruzi) and analyzed the ranking of known and validated antigens obtained from the literature. Using ROC curves we show the performance of our method, and analyze different feature-weighting schemes and their effect on the prioritization of known antigens. Funded by: Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT, Argentina), National Institute of Allergy and Infectious Diseases, NIH (NIH-NIAID, USA).