Package 'SCENT'

Correlation of Connectome And Transcriptome (CCAT)

Description

Estimates signaling entropy via correlation between gene expression and network connectivity.

Usage

CompCCAT(exp.m, ppiA.m)

Arguments

exp.m	Gene expression matrix (genes x cells). Supports dense or sparse.
ppiA.m	PPI network adjacency matrix (genes x genes).

Details

CCAT computes the Pearson correlation between gene expression and network degree (number of interaction partners) for each cell. This approximation is based on the observation that pluripotent cells express hub genes at higher levels.

The method automatically applies log2(x+1) transformation if the input appears to be raw counts (max > 50).

Value

Numeric vector of CCAT scores (one per cell). Higher values indicate higher differentiation potency.

Author(s)

Andrew E Teschendorff, Zaoqu Liu

References

Teschendorff AE, Enver T. Nat Commun. 2017;8:15599.

Examples

## Not run: 
data(net13Jun12.m)
exp <- matrix(rpois(5000 * 100, 5), nrow = 5000)
rownames(exp) <- head(rownames(net13Jun12.m), 5000)
ccat <- CompCCAT(exp, net13Jun12.m)

## End(Not run)

---

Compute Signaling Entropy Rate (SR)

Description

Computes single-cell differentiation potency using the signaling entropy rate on a protein interaction network.

Usage

CompSRana(integ.l, local = FALSE)

Arguments

integ.l	Output from DoIntegPPI, containing expMC and adjMC.
local	Logical. If TRUE, returns normalized local entropies. Default FALSE.

Details

The signaling entropy rate is defined as:

$SR = \sum_j \pi_j S_j / SR_{max}$

where $\pi_j$ is the stationary distribution and $S_j$ is the local entropy at gene j. $SR_{max}$ is the maximum entropy rate of the network, computed from the largest eigenvalue of the adjacency matrix.

Value

List with elements:

SR

Signaling entropy rate (0-1). Higher = higher potency.

inv

Stationary distribution matrix (genes x cells).

locS

Local entropy matrix (genes x cells).

nlocS

Normalized local entropies, or NULL if local = FALSE.

Author(s)

Andrew E Teschendorff, Zaoqu Liu

References

Teschendorff AE, Enver T. Nat Commun. 2017;8:15599.

Examples

## Not run: 
data(net13Jun12.m)
exp <- matrix(rpois(5000 * 50, 5), nrow = 5000)
rownames(exp) <- head(rownames(net13Jun12.m), 5000)
integ <- DoIntegPPI(exp, net13Jun12.m)
sr <- CompSRana(integ)

## End(Not run)

---

Integrate Expression Data with PPI Network

Description

Finds common genes between expression data and PPI network, then extracts the maximally connected subnetwork.

Usage

DoIntegPPI(exp.m, ppiA.m)

Arguments

exp.m	Gene expression matrix (genes x cells). Should be library-size normalized. Raw counts will be log2-transformed automatically.
ppiA.m	PPI network adjacency matrix (genes x genes).

Details

The function:

1. Identifies genes present in both expression data and PPI network

2. Extracts the induced subnetwork

3. Finds the largest connected component

4. Returns matched expression and adjacency matrices

Value

List with elements:

expMC

Expression matrix for genes in the maximal component.

adjMC

Adjacency matrix of the maximal connected component.

Author(s)

Andrew E Teschendorff, Zaoqu Liu

References

Teschendorff AE, Enver T. Nat Commun. 2017;8:15599.

Examples

## Not run: 
data(net13Jun12.m)
exp <- matrix(rpois(5000 * 100, 5), nrow = 5000)
rownames(exp) <- head(rownames(net13Jun12.m), 5000)
integ <- DoIntegPPI(exp, net13Jun12.m)

## End(Not run)

---

Protein-Protein Interaction Network

Description

Protein-protein interaction network adjacency matrix compiled from multiple public databases. Contains 8434 genes.

Usage

net13Jun12.m

Format

A symmetric binary matrix (8434 x 8434):

Rows/Columns

Gene Entrez IDs

Values

1 = interaction, 0 = no interaction

Source

Compiled from public PPI databases (June 2012).

References

Teschendorff AE, Enver T. Nat Commun. 2017;8:15599.

Examples

data(net13Jun12.m)
dim(net13Jun12.m)
sum(net13Jun12.m) / 2  # Number of interactions

---