Skip to contents

scTenifoldKnk

📖 Documentation: https://zaoqu-liu.github.io/scTenifoldKnk/

Overview

scTenifoldKnk is a computational framework for performing in-silico gene knockout experiments using single-cell RNA sequencing (scRNA-seq) data. By leveraging gene regulatory network (GRN) perturbation analysis, scTenifoldKnk enables researchers to predict the transcriptional consequences of gene knockouts without conducting wet-lab experiments.

The method integrates several advanced computational techniques:

  • Principal Component Regression (PCR) for robust GRN inference
  • CP Tensor Decomposition for network denoising and averaging
  • Non-linear Manifold Alignment (NLMA) for network comparison
  • Chi-square Statistics for identifying differentially regulated genes

Methodology 

┌─────────────────────────────────────────────────────────────────────────────┐
│                         scTenifoldKnk Workflow                              │
├─────────────────────────────────────────────────────────────────────────────┤
│                                                                             │
│  ┌──────────┐    ┌──────────┐    ┌──────────┐    ┌──────────┐              │
│  │ scRNA-seq│───▶│ Quality  │───▶│ Network  │───▶│  Tensor  │              │
│  │   Data   │    │ Control  │    │  Build   │    │ Decomp.  │              │
│  └──────────┘    └──────────┘    └──────────┘    └──────────┘              │
│                                                        │                    │
│                                                        ▼                    │
│  ┌──────────┐    ┌──────────┐    ┌──────────┐    ┌──────────┐              │
│  │   Diff   │◀───│ Manifold │◀───│    KO    │◀───│    WT    │              │
│  │Regulation│    │Alignment │    │ Network  │    │ Network  │              │
│  └──────────┘    └──────────┘    └──────────┘    └──────────┘              │
│                                                                             │
└─────────────────────────────────────────────────────────────────────────────┘

The scTenifoldKnk workflow consists of six sequential steps:

  1. Quality Control: Filter low-quality cells based on library size, gene detection rate, and mitochondrial content
  2. Network Construction: Build multiple GRNs from random cell subsamples using principal component regression
  3. Tensor Decomposition: Apply CANDECOMP/PARAFAC (CP) decomposition to denoise and average networks
  4. Virtual Knockout: Simulate gene knockout by setting target gene’s outgoing regulatory edges to zero
  5. Manifold Alignment: Embed wild-type and knockout networks into a shared low-dimensional space
  6. Differential Regulation: Identify genes significantly affected by the virtual knockout using chi-square statistics

Installation 

install.packages("scTenifoldKnk", 
                 repos = c("https://zaoqu-liu.r-universe.dev", 
                          "https://cloud.r-project.org"))

From GitHub 

# install.packages("devtools")
devtools::install_github("Zaoqu-Liu/scTenifoldKnk")

Quick Start 

library(scTenifoldKnk)

# Load example dataset
data_path <- system.file("single-cell/example.csv", package = "scTenifoldKnk")
countMatrix <- as.matrix(read.csv(data_path, row.names = 1))

# Perform virtual knockout analysis
result <- scTenifoldKnk(
  countMatrix = countMatrix,
  gKO = "TargetGene",
  qc_minLSize = 1000,
  nc_nNet = 10,
  nc_nCells = 500,
  td_K = 3
)

# Examine differentially regulated genes
head(result$diffRegulation)

Output Structure

The scTenifoldKnk() function returns a list containing:

Component Description
tensorNetworks$WT Denoised wild-type gene regulatory network
tensorNetworks$KO Perturbed knockout gene regulatory network
manifoldAlignment Joint embedding matrix from NLMA
diffRegulation Data frame of differential regulation statistics

Differential Regulation Table

Column Description
gene Gene identifier
distance Euclidean distance between WT and KO embeddings
Z Z-score (standardized distance)
FC Fold change relative to background
p.value Chi-square test p-value
p.adj Benjamini-Hochberg adjusted p-value

Documentation

Comprehensive tutorials are available at the package website:

Vignette Description
Quick Start Basic usage and example workflow
Algorithm Theory Mathematical foundations and methodology
Visualization Result visualization techniques
Advanced Usage Parameter optimization and customization
Result Interpretation Biological interpretation guidelines

Citation

If you use scTenifoldKnk in your research, please cite:

Osorio D, Zhong Y, Li G, Xu Q, Yang Y, Hillhouse A, Chen J, Davidson L, Tian Y, Chapkin R, Huang J, Cai JJ. scTenifoldKnk: An efficient virtual knockout tool for gene function predictions via single-cell gene regulatory network perturbation. Patterns. 2022;3(3):100434. doi: 10.1016/j.patter.2022.100434

Authors

  • Zaoqu Liu - Maintainer ()
  • Daniel Osorio - Original Author
  • James J. Cai - Principal Investigator

License

This package is licensed under GPL (>= 2).

  • scTenifoldNet - Comparative single-cell GRN analysis
  • Seurat - Comprehensive scRNA-seq analysis toolkit