BRAIN ORGANOIDS AS MODELLING CONDUITS OF HUMAN NEURODIVERSITY: ADVANCED COMPUTATIONAL APPROACHES TO SINGLE CELL TRANSCRIPTOMICS

The establishment of the adult central nervous system (CNS) morphology and physiology results from finely orchestrated developmental processes. The introduction of human brain organoids (BOs), allowed to reach an unprecedented experimental tractability of brain development trajectories, surpassing the limitations of their preceding 2D models and of animal models in recapitulating CNS cytoarchitecture. The advent of single cell (SC) RNA sequencing (scRNA-seq) offered the possibility to pair BOs morpho-functional complexity to its molecular characterization. Despite the recent ameliorations of BOs, they are facing two main technological limitations that prevent accessing their full potential. The first is based on the necessity to increase the sample size of in vitro modelling studies to discover the causal mechanisms of genetic-phenotype associations from population genomics studies. The second is the demand for more granular modelling capabilities, a mandatory requirement to gather insights on region-specific molecular architecture of the brain. To overcome the first technological limitation, we assessed the use of multi-genotype cortical brain organoids (CBOs)-designs as effective scale-up platforms, to study SC phenotypes of human neurodevelopment. I started by benchmarking existing computational tools for SC-genetic demultiplexing when applied to CBOs and developed methods to surpass their performances. Indeed, to accurately deconvolve genetic identities of multiplexed organoids, I developed SCanSNP, to recall genetic identities and a combined call approach that summarises multiple methods agreement. Moreover, we introduced mosaic cortical brain organoids (mCBOs) and compared them to canonical CBOs, while assessing their longitudinal adherence to in vivo corticogenesis. I leveraged this multi-genotype design to explore interindividual variability in neuronal migration, a process that is particularly relevant to cortical expansion and layering. Moreover, to tackle the need for more granular modelling capabilities, I mapped the SC transcriptional diversity of polarized cortical assembloids (polCAs), a new organoid model - originally developed in the Knoblich laboratory- aimed to reproduce the transcriptional diversity across areas of the developing neocortex. First, I assessed polCAs ability to recapitulate transcriptional signatures of the neocortical rostro-caudal axis. Then, I devised a computational framework to extract genes bearing fetal-fronto-temporal gradient expression, consolidating polCAs transcriptional topography in a comparative setting. My work sits at the interface between the experimental and computational improvement of BOs, ensuring a multi-view characterization of newly introduced scalability and modelling opportunities.