Lead Bioinformatician Harvard Medical School, United States
Abstract: The complexity of the mammalian retina arises from the unique combination of extrinsic and intrinsic factors in its development. Several single-cell RNA seq datasets have been generated to study retina and retinal organoids. We propose to investigate human retinal development and the questions of cell fate specification, cell class and cell type heterogeneity, and regulation of cell trajectory by intrinsic and extrinsic factors. We start with the scRNA-seq data for human developing healthy retina and retinal organoids. We focus on the methods of data transformation (Harmony, scVI, ForceAtlas2) and cell-cell interactions analysis (CellChat, Scriabin), cell fate trajectory reconstruction and time transformation (scvelo, scFates, CytoTRACE2). We establish a human fetal retina and human retinal organoids atlas by integrating scRNA-seq datasets into a high-resolution map of retinal development. We use pseudotime with the cell fate trajectory reconstruction, potency and trajectory density methods to deconvolute the transcriptional signal. We demonstrate the application of pseudotime approaches in reconstruction of retinal ganglion cells (RGC) development to unravel the heterogeneity behind cell maturation. We discover hundreds of de novo development-oriented targets from both cell-intrinsic and cell-extrinsic prospectives of retina and RGC maturation. We show how maturation states contribute to forming the environment sufficient for retina development. Our findings confirm developing retina to be a closed and self-supporting system. We demonstrate the comparison of 27 organoids differentiation protocol with such metrics as cell commitment, similarity to native retina, development timepoint correlation, and transplantation probability prediction. We integrate and publish two new tools for the data analysis: a human fetal retina and human retinal organoids atlas. We show how switching from canonical timepoint to pseudotime in studying the cell fate changes the understanding of fate driving genes. Both the atlases can be used as a community tool for analyzing, reference mapping and annotation of the sequencing data. The resulting reference map, publicly available at CellxGene portal, serves as a template for cell differentiation, reprogramming, and transplantation.
Funding Source: NIH/NEI–5U24EY029893-05 (P.B.), NIH/NEI–P30EY003790 (Core Facility Grant), Bright Focus Foundation–G2020231 (P.B.), Gilbert Family Foundation–GFF00 (P.B.), Department of Defense - VRP FTTSA (P.B).
