• Citer

Phylogeographic analysis of the 2018-2020 Ebola virus outbreak in the North-Kivu province of the Democratic Republic of the Congo.

par De Fraipont, Adrien
Président du jury Bogaerts, Philippe
Promoteur Dellicour, Simon
Publication Non publié, 2023-06-12

Mémoire

• Accès en ligne
• Détails
• Contenu
• Statistiques

Résumé :

Context: Epidemics caused by viruses represent one of the greatest health risks facing oursociety. To fight them efficiently, it is essential to develop tools that allow for a betterunderstanding of their complex dynamics. It is the case of phylogeography, a discipline thatuses genetic, spatial, and temporal data of causal agents, viruses in this case, to reconstruct thespatiotemporal history of their dispersal. The Ebola virus epidemic that struck the DemocraticRepublic of the Congo's North Kivu province and its surrounding area from August 2018 toJune 2020 represents an interesting study-case on which to apply this technique, since a notableamount of data has been made available for this outbreak.Goals: This work aims to test methodologies for continuous phylogeographic analysis on thedata of the 2018-2020 EBOV epidemic. Specifically, the goal is to try two different approachesfor the integration of spatial information and assess what possible limitations they couldencounter. We also aim to use the results obtained by those analyses to get an overview of thedynamics of the outbreak and to estimate dispersal statistics, taking into consideration thelimitations observed for the methods.Methods: We performed a continuous phylogeographic analysis using a Bayesian inferenceprocedure on 767 Ebola virus genomic sequences collected between July 2018 and April 2020in three provinces of the Democratic Republic of Congo. In the absence of precise samplinglocations and because the continuous phylogeographic inference requires the assignation ofgeographic coordinates to the samples, we tried two different approaches to retrieve samplingcoordinates. The first exploratory approach that we used simply consisted in drawing randompoints from the health zone of origin of each sample. We then adopted a second approachconsisting in defining homogeneous prior ranges of sampling coordinates delimitated by thepolygon surrounding each health zone of origin of the samples.Results: We observe two limitations regarding the use of continuous phylogeography on thetype of data available in the case of the 2018-2020 outbreak. For the first method, the locationsrandomly drawn inside the health zones cannot be considered good estimates when theadministrative areas are relatively large, since those locations can severely vary from one drawto another. For the second approach, we suspect the creation of an artifact that moves theinferred locations towards the epicenter of the outbreak, leading to an unrealistic reconstructionof the outbreak. Nevertheless, the first analysis still provided results coherent with previousstudies of the outbreak, and the dispersal statistics estimated enable a comparison with the 2014-2016 EBOV Outbreak in Western Africa, showing us that on average, infected individuals wereexploring 2.6 times more surfaces and the virus was dispersing 1.7 times faster during that firstoutbreak.Conclusion: Continuous phylogeography has the potential to provide detailed information onthe course of an outbreak but its application can be limited by the availability of precisesampling locations associated with genomic sequences. To circumvent this issue, we here adoptan approach consisting in integrating sampling uncertainty across the administrative polygonof origin of each sample. However, our related continuous reconstruction resulted in a somehowsuspicious phylogeographic pattern with tip nodes that seem to be artifactually attracted nearthe epicenter of the epidemic. One further perspective would be to apply a third approachaiming at using external information on outbreak records to further constrain the samplinguncertainty within each polygon of origin.