Sourakhata TIRERA

The Amazon is home to one of the greatest diversity of vertebrate and invertebrate species, which are potential reservoirs and vectors of zoonotic diseases. Over the last few decades, this neotropical region has experienced a number of viral epidemics associated with global changes, including both anthropogenic (deforestation, urbanisation, introgression of human populations into intact areas) and climatic (warming, changes in rainfall patterns, drought) modifications. These changes have an impact on the distribution and density of vector and reservoir populations, which can lead to increased contact in interface areas with human populations, resulting in outbreaks.

French Guyana has all the environmental, demographic and socio-economic conditions required for the development of emerging and re-emerging infectious diseases. It is home to a wide variety of flora and fauna, suggesting a high level of viral diversity. Despite this, the region is one of the least studied in terms of viral diversity and associated zoonotic diseases.

Rodents have been described as reservoirs of many human pathogens, with over 60 identified, including at least 20 viruses of interest. This large number is linked to their status as commensal species for humans. In French Guiana, rodents have been described as reservoirs for various viruses. These include the Maripa virus, an aetiological agent of hantaviruses that cause cardiopulmonary syndromes in humans, and two arenaviruses (the LCMV and Patawa viruses) that can cause meningitis and haemorrhagic fevers respectively in the most severe cases.

The aim of this thesis is to characterise the viral diversity hosted by different rodent species in contrasting habitats (primary and disturbed forests, savannah zones and peri-urban areas) and to gain a better understanding of how the environment influences the diversity of the viruses identified. However, making as reliable an inventory as possible of the viral sequences present in a virome can prove complex. Analytical and bioinformatics processing methods can introduce biases into the composition of viromes. To overcome this inventory problem, we have developed a pipeline which, when applied to metagenomic data, enables the detection and assignment of chimeric sequences and a gain in the number of viral sequences identified.

Firstly, an overview of advances in research into zoonotic and vector-borne infectious diseases in French Guiana highlighted the contribution made by the latest analytical tools, based on high-throughput molecular and serological methods, in characterising pathogens, their hosts and vectors over the last decade.

In the second part, viral diversity was characterised in seven rodent species evolving in four types of habitat in order to explore how environmental disturbances influence this diversity. Using a metagenomic approach, 77,767 viral sequences were identified from spleen, kidney and serum samples. These sequences were attributed to 27 viral families known to infect vertebrates, invertebrates, plants and amoebae. Comparisons of viral diversity in species present in at least two habitats showed overall (with one species showing an inverse pattern) a higher diversity in primary forest habitats compared with disturbed habitats. These diversities were also lower in peri-urban areas. High viral richness was also observed in savannah areas. The differences in diversity observed can be explained by the presence of rare viruses, which are generally more frequent in primary forests and savannah habitats. Changes in the ecology and behaviour of rodents in a given habitat, such as changes in diet in disturbed versus undisturbed forests, are likely to be major determinants of viral composition.

In the third part, we discuss the difficulty of exhaustively detecting the viral sequences of a virome by shotgun sequencing. The first step in characterising viral diversity is to produce a reliable inventory, which is currently done by taxonomic assignment of virome sequences. The analytical (concentration of viral particles, extraction method, random amplifications, sequencing method) and bioinformatic (for the production of contigs representing a virome) approaches introduce biases such as over-representation of certain genomes and the formation of chimeric sequences composed of different organisms. To address this issue, we have developed a pipeline called 'd-chimer' that enables taxonomic assignment of post-assembly sequences, optimising viral detection in viromics studies, including in chimeric contigs. This approach is based on two steps. Firstly, a filter takes into account the possibility of the presence of chimeric sequences in BLAST output, illustrated by different fractions of the query aligning with different reference sequences. Then, considering that not all parts of sequences are detected in this process, the uncovered areas are cut, stored and used as input for a new BLAST - filter cycle, in a recursive manner (recycling). Using d-chimer in two successive stages of BLASTn and BLASTx significantly reduces overall computing time, while enabling better detection of viral fragments, with a positive impact on the richness detected compared with the 'best-hit' approach. We applied d-chimer to assembled data from bird and rodent viruses. Compared with a one-step best-hit approach, both datasets show a significant increase in the number of viral contigs and contig fragments assigned. This increase in the number of viral sequences also revealed an increase in viral richness at species and genus level and, to a lesser extent, at family level.

All the results obtained during this thesis show that environmental components are key factors in viral diversity. Combined with the development of high-throughput molecular detection tools, analyses of viral diversity in animal reservoirs should enable us to better study the ecology of viruses and their determinants. Comparing viral metagenomic data obtained from animal reservoirs, vectors and humans should shed light on the links with zoonoses in French Guiana, and elsewhere in Amazonia, in ecosystems whose high viral diversity remains to be understood.

Amazonia hosts one of the greatest diversity of vertebrate and invertebrate species, potentially reservoirs and vectors of zoonotic diseases. In the last decades, this neotropical region has experienced numerous viral epidemics associated with global changes, including both anthropogenic (deforestation, urbanisation, introgression of human populations into intact areas) and climatic modifications (warming, changes in rains regime, drought). These changes impact the distribution and densities of vector and reservoir populations, which can lead to increased contact at the interface areas with human populations and thus lead to emergence.

French Guiana harbours all the environmental, demographic, and socio-economic conditions that are favourable to the emergence and re-emergences of infectious diseases. This French department hosts a great diversity of flora and fauna, which may reflect a significant viral diversity. Nevertheless, the region is one of the least studied in terms of viral diversity and associated zoonotic diseases.

Rodents have been described as reservoirs of numerous human pathogens (more than 60 identified), including at least 20 viruses of interest. This large number is associated with their status as commensal species of humans. In French Guiana, rodents have been described as reservoirs of various viruses. Among them, we can mention the Maripa virus, an etiological agent of hantaviruses that induce cardiopulmonary syndromes in humans, and two arenaviruses (LCMV and Patawa viruses) that can respectively cause meningitis and haemorrhagic fevers in the most severe cases.

The objective of this thesis is to characterise the viral diversity hosted in different rodent species present in contrasting habitats (primary and disturbed forests, savannah, and peri-urban areas) and to better understand how the environment influences the diversity of the identified viruses. However, making the most reliable inventory of the viral sequences present in a virome can be complex. Indeed, analytical methods and bioinformatics processing can introduce biases in the composition of viromes. In order to overcome this inventory problem, we have developed a pipeline, which, when applied to metagenomic data, allows the detection and assignment of chimeric sequences and a gain in the number of viral sequences identified.

Firstly, a synthesis of the most recent research advances on zoonotic and vector-borne infectious diseases in French Guiana highlighted the contribution of the latest analytical tools, based on high-throughput molecular and serological methods, in the characterisation of pathogens, their hosts and vectors over the last decade.

Second, viral diversity was characterised in seven rodent species evolving in four types of habitats in order to explore how environmental perturbations influence this diversity. Using a metagenomic approach, 77,767 viral sequences were identified from spleen, kidney and serum samples. These sequences were assigned to 27 viral families known to infect vertebrates, invertebrates, plants and amoeba. Comparisons of viral diversity in species present in at least two habitats showed overall a higher diversity in primary forest habitats compared to disturbed habitats. These diversities are also lower in peri-urban areas. High viral richness is also observed in savannah areas. The differences in diversity observed can be explained by the presence of rare viruses that are generally more frequent in primary forest and savannah habitats. In addition, changes in the ecology and behaviour of rodents in a given habitat, such as changes in diet in disturbed versus undisturbed forests, should be major determinants of viral composition.

Third, we focused on the difficulty to detect exhaustively the viral sequences of a virome by shotgun sequencing. The first step in characterising viral diversity is to produce a reliable inventory, which is currently done by taxonomic assignment of virome sequences. However, the analytical (concentration of viral particles, extraction method, random amplifications, sequencing method) and bioinformatics (for the production of contigs representing a virome) approaches introduce biases such as an over-representation of certain genomes and the formation of chimeric sequences composed of different organisms. To address this issue, we have developed a pipeline called "d-chimer" that allows taxonomic assignment of post-assembly sequences to optimise viral detection in viromics studies, including in chimeric contigs. This approach is based on two steps. First, a filter that takes into account the likelihood of the presence of chimeric sequences in the BLAST outputs expressed by different fractions of the query aligning with different reference sequences. Then, considering that not all parts of sequences are detected in this process at once, the uncovered areas are cut, stored and used as input for a new BLAST - filter cycle, in a recursive manner (recycling). The use of d-chimer in two successive steps of BLASTn and BLASTx significantly reduces the overall computational time, while allowing a better detection of viral fragments with a positive impact on richness compared to the "best-hit" approach. We applied d-chimer to assembled bird and rodent viral data. Compared to a one-step " best-hit " approach, both datasets show a significant increase in the number of viral contigs and contig fragments assigned. This gain in viral sequence numbers also revealed an increase in viral richness at the species and genus level and, to a lesser extent, at the family level.

All the results obtained during this thesis show that environmental components are key factors in viral diversity. Combined with the development of high-throughput molecular detection tools, the analysis of viral diversity in animal reservoirs should allow us to better study the ecology of viruses and their determinants. The comparison of viral metagenomic data obtained in animal reservoirs, vectors and humans should shed light on the links with zoonoses in French Guiana and elsewhere in Amazonian ecosystems where a significant part of viral diversity remains to be discovered and understood.