Lethal COVID-19 Single Nuclei RNA Sequencing Analysis

This repository contains a comprehensive analysis of single-nuclei RNA sequencing (snRNA-seq) data from patients who succumbed to COVID-19. The study aims to elucidate the molecular and cellular mechanisms underlying severe outcomes in COVID-19 by leveraging advanced computational techniques.

Repository Structure

The analysis is organized into several key stages, each documented in individual Jupyter notebooks:

1. Quality Control: jupyter notebook

   • Overview: Initial assessment and filtering of raw snRNA-seq data to ensure high-quality inputs for downstream analysis.
   • Key Steps:
     • Removal of low-quality nuclei and potential doublets.
     • Outlier detection using Median Absolute Deviation.
     • Assessment of sequencing depth and mitochondrial gene content.

2. Normalization: jupyter notebook

   • Overview: Standardization of gene expression measurements across nuclei to account for technical variability.
   • Key Steps:
     • Comparison of various normalization and transformation methods.
     • Application of scaling factors to normalize counts.
     • Logarithmic transformation to stabilize variance.
     • Scran size factor-based normalization.
     • Pearson's residuals transformation.

3. Feature Selection: jupyter notebook

   • Overview: Identification of highly variable genes that capture significant biological signals.
   • Key Steps:
     • Comparison of various feature selection methods.
     • Scry-based deviant gene determination.
     • Seurat highly variable gene identification.
     • Pearson's residual-based highly variable gene determination.
     • Selection of genes for downstream analyses based on variability thresholds.

4. Dimensionality Reduction: jupyter notebook

   • Overview: Reduction of data complexity to facilitate visualization and clustering.
   • Key Steps:
     • Principal Component Analysis (PCA) to identify major axes of variation.
     • Visualization using Uniform Manifold Approximation and Projection (UMAP) and t-Distributed Stochastic Neighbor Embedding (t-SNE).

5. Integration: (In Progress)

   • Overview: Combining data from multiple samples to create a unified dataset.
   • Key Steps:
     • Alignment of datasets to correct for batch effects.
     • Merging of datasets for comprehensive analysis.

6. Quality Assessment: (In Progress)

   • Overview: Evaluation of the integrated dataset to ensure data integrity and consistency.
   • Key Steps:
     • Assessment of integration success.
     • Identification of any remaining technical artifacts.

7. Cell Annotation: (In Progress)

   • Overview: Classification of nuclei into distinct cell types based on expression profiles.
   • Key Steps:
     • Assignment of cell type identities using known marker genes.
     • Experimentation with reference-based cell annotation.

8. Differential Gene Expression: (In Progress)

   • Overview: Identification of genes with varying expression levels between conditions or cell types.
   • Key Steps:
     • Statistical testing to detect differentially expressed genes.
     • Functional enrichment analysis to interpret biological significance.

9. Trajectory Analysis: (In Progress)

   • Overview: Reconstruction of dynamic processes and lineage relationships among cells.
   • Key Steps:
     • Pseudotime analysis to infer developmental trajectories.
     • RNA velocity analysis.