Exploring biological signals: from pathway visualization to spatial transcriptomics.

The present Ph.D. thesis describes different projects concerning two main topics: visualization of biological networks and the identification of spatially variable genes (SVGs), which are genes that have a spatial pattern of expression. They can be identified from Spatially Resolved Transcriptomics (SRT) data, as it preserves spatial information of the tissue's cells. Although those projects are involved in two different topics, they share a common theme, the visualization of biological data. It is essential to access network information and to visualize a clear spatial pattern of expression of the SVGs identified, which can be related to the spatial changes of the tissue under study. The main project of the thesis is itGraph, which concerns a novel web tool to explore pathways of interest with three different network perspectives. To enhance the visualization of the provided biological networks, it integrates features that are still lacking in other similar software, which also improve the tool's utility and the user experience. For this project, I developed a pipeline to import pathways from the R package graphite (Sales et al. 2012) and created the above-mentioned three network perspectives. The pipeline also includes a strategy to pre-compute their layout before the actual visualization. Furthermore, I optimized all the scripts, the server (backend), the database, and the client (frontend), including the graphical interface of the entire tool. The second project concerning network visualization is called MyoData, which was completed and published in the “Computational and Structural Biotechnology Journal”. It is a comprehensive and integrated resource for single myofiber and nucleus miRNA:lncRNA:mRNA coregulatory networks, also evaluating their impact in relation to known pathways such as those present in the KEGG collection. It integrates a minimal version of the network visualization tool used in itGraph. The aim of the other two projects regards the identification of SVGs. At first, I describe SpatialDE, an R package wrapping of the SpatialDE Python method. It was created to respond to a challenge proposed at the EuroBioc2020 Conference and was published on Bioconductor in October 2021. The last project is called VoyageR. The objective was to conduct a benchmark of computational methods to identify SVGs. This kind of study can be an important guide that can help users to choose the best suitable method for their use. Moreover, new methods for performing this type of analysis continue to be published. For this reason, my project not only benchmarks a large number of published pipelines but was designed to be extensible in order to simplify the addition of further methods.