Projects

Expression prediction

Surprisingly, snoRNAs encoded within the same host gene can have abundance levels that greatly differ. Furthermore, only one third of all known snoRNAs in human are expressed in physiological conditions. To better understand these observations, we combine the use of machine learning and large scale expression datasets in several healthy human tissues (brain, liver, ovary, testis, etc.). Based on these data, we build various predictive models and interpret their decisions, which will identify the main factors that regulate the expression of snoRNAs in human.

Interaction prediction

SnoRNAs recognize their RNA targets by sequence complementarity in order to guide their chemical modification, such as methylation. The current tools for predicting snoRNA interactions are very limited, so we harnessed the datasets from high throughput RNA-RNA interactions identification methodologies to train a machine learning model to recognize snoRNA targets. This tool will facilitate and deepen the study of snoRNAs.

snoRNA and Ovarian Cancer

Over the last couple years snoRNAs have been discovered to play an important role in multiple different cancers. Their expression can be either upregulated or downregulated but the causality of this dysregulation is still unknown. Through TGIRT-Seq of 3 low grade and 3 high grade  ovarian cancers, we have discovered a selective group of snoRNAs to be dysregulated between those two cancer types. The goal of this project is to establish the role of these snoRNAs in tumour aggressiveness by knocking down their expression and studying the phenotypic effect on model ovarian cancer cell lines.

Biomarker and gliomas

Biomarkers are measurable characteristics that describe a specific biological state. The use of these biomarkers represents a significant factor for better personalized patient management. Thanks to the democratization of RNA sequencing and the availability of the data generated in public databases such as the TCGA, the objective of this project is to determine all the genomic and genetic factors that could negatively or positively influence the survival of patients with the most common and aggressive cancer, the glioma.

Single cell and Rheumatoid Arthritis

Rheumatoid Arthritis (RA) is an inflammatory autoimmune disease that affects 1% of the world's population. An intriguing aspect of this disease is that not all patients respond to the same treatments despite sharing the same symptoms. Our hypothesis is that under the common diagnosis of RA could hide a plurality of genetic dysfunctions resulting in the same symptoms.

To explore this hypothesis, single-cell RNA-seq data is extracted from PBMC samples from patients diagnosed with RA but still untreated. The analysis of these single-cell data allows us to study the RNA expression of these cells with great precision by targeting certain cell types for example. The main objective of the project is to compare samples from several patients with RA in order to characterise different immune cell expression profiles and thus define immune endophenotypes.

Embedded snoRNAs and splicing regulation

In 2018, a study showed that a snoRNA, SNORD86, was able to regulate the splicing pattern of its host gene, modulating its expression through a feedback loop. Using high-throughput data from the literature that maps RNA-RNA interactions in cells, we found that a large proportion of snoRNAs interact with their host gene and that several show some evidence that these interactions could affect the profile of splicing of their host gene. In order to better understand this phenomenon, this project aims to characterize these different interactions in several sets of transcriptomic data from the literature of different types to create an interaction network that will allow us to target the most promising candidates.

Creation of bioinformatics tools and websever

We have great expertise in the field of snoRNAs and during our analyzes, we realized that the available tools and databases were not well suited to the study of snoRNA. In an attempt to improve the situation and to share our knowledge with the scientific community, we have developed certain tools and web servers such as CoCo and snoDB (see the tools section) to facilitate the study of snoRNA. Everything is mainly developed with the Python language, but many other languages/technologies are also used such as Bash, PostgreSQL, HTML, CSS, JavaScript, Docker, Snakemake and R. We are currently developing and improving certain tools and web servers on snoRNAs, but also on other topics such as alternative splicing.