CE_08C: Statistical and machine learning methods for single-cell and spatial transcriptomics data analysis (Added Fee)

Recent developments in single-cell and spatial transcriptomics (ST) technologies have enabled scientists to get an integrated understanding of cell types and states and the spatial relationships of different cells in their morphological context. Applications of these technologies in diverse tissues and diseases have transformed our views of transcriptional complexity. With the advantages of single-cell RNA-seq (scRNA-seq) and ST come statistical challenges that are just beginning to be addressed. In this course, we will review the computational and statistical methods available for the analysis of scRNA-seq and ST data. Module 1: Quality control and normalization in scRNA-seq We will review background and technology development for scRNA-seq. We will also discuss important steps in quality control in scRNA-seq data analysis. At the end of this module, we will review the most commonly used methods for scRNA-seq normalization. Module 2: Gene expression denoising and imputation In this module, we will review commonly used statistical methods for gene expression denoising and imputation that aim to recover the true expression of genes in each individual cell and improve gene expression quality. Module 3: Clustering and batch effects removal In this module, we will review commonly used tools for dimension reduction and clustering for scRNA-seq data analysis. We will also discuss how to remove batch effects in clustering as well batch effect removal in the original gene expression space. Module 4: Pseudotemporal ordering and trajectory reconstruction While scRNA-seq experiments provide for such profiling, they are limited to snapshot experiments, meaning that continuous monitoring of genome-wide gene expression in individual cells over time is not possible. In this module, we will review commonly used methods for pseudotemporal ordering and trajectory reconstruction from scRNA-seq data. Module 5: Spatial domain detection In ST studies, an important step is to cluster spots and identify spatial domains, i.e., regions that are spatially coherent in both gene expression and histology. In this module, we will review recently developed tools for spatial domain detection and compare the different approaches they used to integrate the gene expression with spatial location and histology information. Module 6: Spatially variable gene selection Spatially variable genes (SVGs) are genes that show spatial expression variation across a tissue section. SVG selection is an important first step towards characterizing the spatial and functional organization of complex tissues. In this module, we will review methods for SVG detection and compare their advantages and drawbacks. Module 7: Integration of scRNA-seq and ST data Given the complementary information provided by scRNA-seq and ST, one can use statistical approaches to integrate these two data types to infer the spatial locations of different cell types in a tissue. In this module, we will review methods that model scRNA-seq and ST data jointly. Module 8: Predicting high-resolution gene expression in ST In sequencing-based ST, gene expression is only measured in a few thousand spots. Since gene expression in neighboring spots is correlated, it is possible to borrow information from neighboring spots to increase gene expression resolution and fill in the unmeasured area. In this module, we will review newly developed methods for gene expression enhancement in ST.