Coursera Spring Sale
40% Off Coursera Plus Annual!
Grab it
Learn about two novel statistical methods that leverage permutation-based techniques to enhance the rigor of single-cell genomics data analysis in this comprehensive seminar from the Broad Institute's Models, Inference and Algorithms series. Discover scDEED, a method that addresses the challenge of evaluating the reliability of two-dimensional embeddings produced by visualization methods like t-SNE and UMAP by calculating reliability scores for each cell embedding and comparing consistency between neighbors in embedding space versus pre-embedding space. Explore how scDEED flags dubious cells with low reliability scores while identifying trustworthy ones, and provides guidance for optimizing hyperparameters to minimize dubious embeddings and improve visualization reliability across datasets. Examine mcRigor, a statistical method designed to enhance metacell partitioning in single-cell RNA-seq and ATAC-seq data analysis by introducing a feature-correlation-based statistic to measure heterogeneity within metacells and identify dubious metacells composed of heterogeneous single cells. Understand how mcRigor optimizes metacell partitioning algorithm hyperparameters to enhance downstream analysis reliability and enables benchmarking to select the most suitable partitioning algorithm for specific datasets. Gain insights into the practical applications of these methods for uncovering differential gene co-expression modules, enhancer-gene associations, and gene temporal expression patterns while ensuring more accurate and reproducible results in complex cellular process analysis. Access accompanying resources including GitHub repositories and published research papers to implement these methods in your own genomics research workflows.
