Conferences

Metabarcoding has become a powerful approach and it is now widely used to study biodiversity from environmental samples. However, it is still prone to some pitfalls. The experimental steps consist of extracting DNA from environmental samples, amplifying a genetic marker by PCR, and sequencing the amplicons. The sequences obtained by high throughput sequencing should be filtered to eliminate occurrences due to experimental errors (false positives), but keeping correct sequences, even if weakly amplified (avoid false negatives). The second step is a taxonomic assignment of validated sequences. In this talk, I will concentrate on the potential problems of metabarcoding and review possible solutions to obtain clean datasets, comparable among studies. Notably, the co-design of experimental (wet lab) protocol and the sequence analyses is a critical point, where mock community samples, negative controls, and replicates should be key parts of the analyses. This co-design helps to find a balance between false positive and false negative occurrences, which is a key point in producing reliable datasets for statistical analyses. The choice of the genetic marker is another key point that depends essentially on the genetic variability of the marker, the reproducibility of the amplification and on the availability of reference sequences for taxonomic assignments.

In the context of global warming, drought, and low soil fertility, the scientific community has been looking at ways to transform agricultural practices to make them more sustainable. Researchers have been analysing the interaction between plants and their root-associated microbiota to decipher the structures and functions of soil communities and to identify bacterial strains best suited for plant growth, plant health and drought resistance. In this context, bacterial biofilms have been recognized to improve soil structure and water retention. Biofilms polymers mainly include exopolysaccharides (EPS) synthesized by the bacteria to interact with their environment and protect the other populations of soil microbiota. Only a few EPS structures have been characterized to date given that such characterizations require complex/costly experiments and that EPS structures are extremely diverse, even across strains of the same species. With the improvements in genome sequencing, it is timely to conduct comparative genomics studies explaining the variation in EPS structures by the rearrangement of their encoding genomic regions. In this work, our efforts were focused on EPS-producing bacteria isolated from agricultural fields. More specifically, we analyzed 11 strains of the sole Rhizobium alamii species, known to encode 4 to 5 different EPS structures based on preliminary 2D NMR on proton and carbon. On the wet-lab side, we confirmed the EPS structures of all R. alamii strains through HPLC mass spectrometry analyses. On the in-silico side, we first identified the genetic determinant of EPS synthesis by in-depth literature reviewing. We then sequenced their genomes using Oxford Nanopore technologies and proceed to detailed genome annotation using Pfam for transporters, decorators and precursors, and CAZy for polysaccharide synthesizing enzymes. Our main contribution was the development of an algorithm integrating protein annotation and gene proximity that results in the prediction of the genomic regions with a score based on their content to quickly sort candidates for EPS synthesis. These predictions across the 11 different strains producing either similar or different EPS, allowed us to show a direct correlation between the content and organization of gene clusters and the EPS composition, and to attribute a putative function to some specific missing/extra genes. We finally tested our predictor on 88 genomes from Rhizobium and Sinorhizobium strains to determine the accuracy of this method and used Jaccard distance to regroup every genomic region by similarity. These comparisons enabled us to regroup strains by EPS structures and validate our algorithm.

Soil is a non-renewable resource with a crucial function for life taking part in essential processes such as nutrient, water, carbon and nitrogen cycles and even it can transfer mass and energy between spheres. In addition, soil is vital for food production and its conservation has become one of the main political responsibilities, however, this task can be complicated given the complexity of soil which includes interactions between plants, roots, animals (as earthworms) or even microorganisms. In recent years, the interest in microbial communities in environment systems is increasing due to the key role they have in the environment, for example, microorganisms play a key role in carbon and nitrogen cycle, helping to capture atmospheric CO2 or even atmospheric N2 contributing to increment nitrogen in soil which could be used by plants avoiding the overuse of fertilizers. The study of microbial communities is a challenge towed to the fact that around of 99% of microorganisms cannot be cultivated in a laboratory but the development of molecular technologies such as shotgun or amplicon sequencing make this challenge possible since they can obtain molecular information of all the microorganisms present in the environmental sample. This massive sequencing generates a lot of information which has to be processed though bioinformatic pipelines which will determine the type of microorganisms present in the sample but also their abundance. In addition, several algorithms have been developed with the aim to discover the potential functionality of those microorganisms such as Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) or FUNguild which use the taxonomic assignments and all the information contained in large databases to find out either a gene expression or guild. In this doctoral thesis, the amplification technique with different bioinformatic approaches were used to study the effect of sustainable agriculture on microbial communities, for that two agrarian system were employed 1) the application of organic amendment instead of mineral fertilizers and 2) the association of melon/cowpea in different patterns instead of the traditional monocropping system. The bioinformatic analysis included the general study of microbial community (alpha and beta diversity), the study of the most abundant taxa and their relationships with soil properties. In addition, a biomarker detection algorithm was also employed to detect which microorganisms had been mostly affected by the agronomic system. On the other hand, to deepen in the effects on bacterial communities, 1) the PICRUSt algorithm was employed to evaluate the effects on geneexpression of some biogeochemical cycles, 2) a co-occurrence networks analysis was carried out to evaluate the interaction between the microorganisms, 3) rare and abundant taxa were studied and 4) a random forest machine learning was employed to determine the importance of soil microorganisms.

Genomic advances have the potential to greatly benefit both research and industry, and online tools and databases are key to making the most of this potential. In plants, these resources are fundamental in the contribution to crop improvement, whether achieved through selective breeding or biotechnological strategies. However, acquiring knowledge about a plant's genome and its genes can be challenging, especially for the economically important but non-model species like Vitis vinifera, also known as grapevine. Research communities studying model organisms have established a variety of online tools and databases serving as repositories of genomic and gene function data. The lack of these types of resources for non-model plants as grapevine drives the need for a centralized, standardized collection of assets within the grapevine community. Despite the efforts of the community and projects like INTEGRAPE to standardize data, the community still struggles with the dispersed resources that are currently available. The GRAPEDIA portal aims to centralize, interconnect, and showcase these resources in a single platform that offers customized solutions for the entire grapevine community, including researchers, breeders, and the wine industry. The target audience for this portal is the entire scientific community working on grapevine or using it as a model plant, as well as the private sector involved in viticulture research and development.

The environmental impact of (scientific) computing, computational biology in particular, is a growing concern in light of the urgency of the climate crisis, and there is widespread interest in the research community; so what can we all do about it? Tackling this issue and making it easier for scientists to engage with sustainable computing is what motivated the Green Algorithms project. We will discuss what we learned along the way, how to estimate the impact of our work and dive into the carbon footprint of common bioinformatic analyses. We will also highlight the levers scientists and institutions have to make their research more sustainable, debate the ethical implications of these environmental costs and examine what is still needed moving forward.

Population genetics is a research field where bioinformatics has a key role. The first part of the presentation highlights the need to develop flexible standard workflows to generate reliable data for non-bioinformatics specialists when processing raw sequencing data. The second part discusses the need to develop robust software for population genetic analyses for non-statisticians and non-bioinformaticians in order to obtain reliable results.