CSN-Literature-Review

A. Overview of single-cell RNA sequencing (scRNA-seq)
1. High-throughput technology for measuring gene expression at single-cell resolution
2. Importance in studying cellular heterogeneity and identifying distinct cell states
B. Challenges in scRNA-seq data analysis
3. High technical noise and low coverage compared to bulk RNA-seq
4. Need for advanced computational methods to uncover biological insights
C. Objective of the review
- To explore how Cell-Specific Networks (CSN) and Conditional Cell-Specific Networks (c-CSN) address these challenges

A. Gene expression-based methods
1. Principal Component Analysis (PCA)
2. Non-negative Matrix Factorization (NMF)
3. t-distributed Stochastic Neighbor Embedding (t-SNE)
B. Clustering methods
4. Hierarchical clustering
5. K-means clustering
6. SNN-Cliq and other advanced methods
C. Trajectory inference methods
7. Monocle
8. TSCAN
9. DPT
D. Limitations of traditional methods
10. Inability to capture cell type-specific gene interactions
11. Lack of a network perspective in understanding gene regulation

A. Definition and rationale
- Transforming gene expression data into gene association networks for individual cells
B. Methodological approach
1. Constructing CSNs by identifying statistically dependent gene pairs in each cell
2. Applying a statistical model to evaluate gene dependency
C. Advances and contributions
3. Single-cell resolution for gene-gene interaction analysis
4. Identification of key regulatory genes and network modules
5. Enabling cell clustering and pseudo-trajectory analysis with Network Degree Matrix (NDM)

A. Uncovering novel biomarkers and regulatory pathways
1. Identification of hidden subpopulations in various datasets
2. Revealing differential gene associations and network features in distinct cell types
B. Integration of CSN tools into existing scRNA-seq analysis approaches
3. Compatibility with PCA, t-SNE, and advanced clustering algorithms
C. Case studies and experimental validations
4. Potential for drug-target identification in cancer (Dai et al.)
5. Analysis of various cell types including macrophages, embryonic neurons, and endothelial cells

A. Addressing the limitations of CSN
- Overestimation problem due to including both direct and indirect gene associations
B. Concept of conditional independence
1. Filtering out indirect associations by conditioning on third genes
2. Construction of sparser and more accurate direct gene association networks
C. Methodology of c-CSN
3. Estimation of conditional probability distributions
4. Identification of direct gene-gene interactions using the three-dimensional gene expression data
5. Deriving Conditional Network Degree Matrix (CNDM) for downstream analyses

A. Superior performance in cell type separation and clustering
1. Enhanced ability to distinguish different cell types in low-dimensional space
B. Capturing network dynamics during cellular differentiation
2. Illustration of networks dynamically changing at various developmental stages
C. Estimating cellular differentiation potency using Network Flow Entropy (NFE)
3. Predictive power in quantifying the differentiation state of individual cells
D. Extending to single bulk RNA-seq data analysis
4. Application to single-sample networks and uncovering disease-related biological connections

A. Computational challenges
1. Increased complexity and need for parallel computing to handle large datasets
B. Non-causal nature of inferred gene associations
2. Future research on incorporating causality and directional gene interactions
C. Need for experimental validation
3. Future studies to validate computational findings with biological experiments

Summary of contributions from CSN and c-CSN methodologies

Importance in advancing scRNA-seq data analysis and understanding complex cellular biology