a Web application for new multidisiciplinary partnerships

MOTS-CLES : Interactivity ; web application ; multi-omics ; R / Shiny

The increasing availability of multi-omics data represents an ideal occasion to study in detail molecular variability that underlies important phenotypes and the interaction between the different levels of biological complexity. Integration, analysis and interpretation of these heterogenous data on different levels represents, however, an important multi-disciplinary challenge that requires close partnerships between biologists, bioinformaticians and biostatisticians.

With this in mind, the "open-source" R software has become an essential tool for the analysis of omics data, its application ranging from ready-to-use existing software packages (e.g. code librairies written by the active user community) to de novo coding of new algorithms and methods. The majority of multi-omics analyses is necessarily situated in a grey zone between these two extremes, where the existing methods should be adapted and extended to allow full exploitation of the data. At the same time, recent progress of the Shiny software (https://shiny.rstudio.com/) now allows the creation of Web interfaces directly from R without specialized Web development knowledge, opening the door to an explorable, interactive and user-friendly way to communicate data and analysis resuts between partners and ultimately end-users.

During an AgreenSkills + mobility at the University of Wisconsin-Milwaukee in the United States (2017-2019), my colleagues and I showed how an interactive web R/Shiny application could considerably improve a collaboration based on multi-omics data analysis. This study used high-throughput coupled gene expression (seqRNA) and activator assays (seqATAC) over a period of time and evolution to characterize the dynamic mechanisms governing gene regulatory programs during central nervous system axon regeneration (Dhara et al., 2019).To facilitate the exploration and functional interpretation of multi-omics data entirely treated on temporal transcription networks, we created an interactive Web application called Rosetta Regeneration (http://ls-shiny-prod.uwm.edu/rosetta ; Rau et al., 2019).

The Rosetta Regeneration application first served as a complement to our internal collaboration by allowing our wet-lab collaborators to explore the analysis results without any programming knowledge. However, the addition of other functionalities to the application indicated that its use could be way beyond that of the original study. In particular, Rosetta Regeneration allows users to select lists of integrated genes or to enter personnalized lists in one of ten organisms supported to (1) visualize the temporal expression trends in clusters; (2) to identify proximal and distal regions of accessible chromatin to speed up downstream pattern analysis; and (3) to describe the ontological categories of enriched functional genes. By allowing simple querying of this rich data without extensive bioinformatics expertise, Rosetta Regeneration is widely useful for an in-depth study of time-dependent regulation during regeneration in zebrafish and for hypothesis generation in other organisms.

With the large-scale generalization of multi-omics data, providing tools that allow to interrogate and explore data without having advanced knowledge of coding is necessary. Rosetta Regeneration may serve as a model for other improved multi-disciplinary collaborations at INRAE. Hosting and deploying R/Shiny web applications online requires dedicated technical solutions. Different teams within the institute are currently exploring their feasibility at INRAE. In the meantime, even internal collaborations can benefit from useful information by taking advantage of interactive analysis tools deployed locally for the analysis of omics data.