Abstract :
[en] Plant viruses are a major cause of crop losses and decreased agricultural productivity worldwide. Rapid and accurate detection of plant viruses is essential for the implementation of effective control measures. Traditional methods of plant virus detection, such as serological and molecular assays, often present very good performance criteria but they are targeted, and they don’t detect new viruses or divergent strains of known viruses.
Overall, developing and validating innovative sequencing tools for fast and efficient detection of plant viruses gained a lot of leverage. Indeed, high throughput sequencing (HTS) tests followed by bioinformatic analyses can detect several viruses at once (including novel ones) and then characterise their genomes. This very high inclusivity allows better monitoring of agricultural pest presence than traditional methods. In addition, the sensitivity of HTS viral detection is theoretically higher than molecular and serological tests, meaning that low-level infection can be traced more efficiently.
HTS tests have several drawbacks: the price, the high technical requirements and the cross-contamination of sequences between samples nevertheless. The cost of viral detection by sequencing is higher than traditional methods, but the cost gap is reducing over time as HTS is more and more affordable. More technical skills are required for sequencing and analysis of a sample for virus detection, but the laboratory and bioinformatic protocols are becoming simpler and easier to learn and apply. Cross-contamination between samples is a recurrent phenomenon that is challenging the operational activities of laboratories aiming to detect plant pests. The high sensitivity of HTS has a drawback as it means that cross-contamination is an even more pressing issue than with traditional methods.
Cross-contamination is probably one of the main issues when using HTS for viral detection. Indeed, if an unexpected genetic material transfer happens between two samples in the laboratory, one virus can be sequenced in the other sample. Since sequencing sensitivity is high, HTS is more prone to detect this cross-contaminating virus. That may lead to a false positive virus detection (as it is really in the bioinformatic data) while it was not present in the plant. The specificities of HTS technologies (high sensitivity, high inclusivity but with the complexity of laboratory and bioinformatics steps) make their validation difficult compared to traditional tests. Therefore, this thesis describes the side-by-side comparison between traditional tests and HTS technologies for virus indexing of Musa germplasm collection. In addition, an alien control (a specific type of external control) has been used for the first time to
II
monitor cross-contamination in HTS. In addition, a newly described alien-based filter algorithm, called Cont-ID, has been developed and applied to find the most appropriate limit of detection that should be applied for accurate virus detection taking into account the risk of false negatives and false positives. That way, the detection prediction's confidence can be high enough to be considered for its use in plant virus diagnosis.
As written above, HTS technologies can also characterise the genome of the detected viruses. Through variant analysis, the different virus variants can be highlighted. A performance testing was conducted to better understand the difficulties and therefore improve the variants' characterisation.
This thesis has therefore addressed several drawbacks limiting potentially the use of HTS technologies for plant virus detection and genome characterisation. It has delivered several milestones to contribute to these technologies' wider and more reliable applications for plant virus detection. Overall, it has reinforced its high potential for improving the control and management of plant virus diseases.
