Type: Package
Title: Computing Weighted Topological Overlaps (wTO) & Consensus wTO Network
Version: 2.1
Author: Deisy Morselli Gysi, Andre Voigt, Tiago Miranda Fragoso, Eivind Almaas and Katja Nowick.
Maintainer: Deisy Morselli Gysi <deisy.ccnr@gmail.com>
Description: Computes the Weighted Topological Overlap with positive and negative signs (wTO) networks given a data frame containing the mRNA count/ expression/ abundance per sample, and a vector containing the interested nodes of interaction (a subset of the elements of the full data frame). It also computes the cut-off threshold or p-value based on the individuals bootstrap or the values reshuffle per individual. It also allows the construction of a consensus network, based on multiple wTO networks. The package includes a visualization tool for the networks. More about the methodology can be found at <doi:10.1186/s12859-018-2351-7>.
License: GPL-2
LazyData: TRUE
Imports: data.table, igraph, magrittr, plyr, parallel, som, visNetwork, reshape2, Rfast, HiClimR, methods
Suggests: knitr, rmarkdown
RoxygenNote: 7.3.1
Encoding: UTF-8
NeedsCompilation: no
Packaged: 2024-07-04 19:54:14 UTC; deisygysi
Repository: CRAN
Date/Publication: 2024-07-04 22:40:01 UTC

CorrelationOverlap

Description

This function computes the correlation between Nodes and the Overlapping Nodes of interest.

Usage

CorrelationOverlap(Data, Overlap, method)

Arguments

Data

data.frame containing the expression data. Nodes on the Rows, Individuals on the Columns. Don't forget to give the names to the Nodes and to the Individuals. Nodes must have the row.names() with the Node Name.

Overlap

A vector containg the names of the Nodes of interest.

method

Spearman ("s", "spearman") or Pearson ("p", "pearson") correlation

Author(s)

Deisy Morselli Gysi <deisy at bioinf.uni-leipzig.de>

Cut.off

Description

Computes the cutoff threshold using the descriptive values generated by the bootstrap, also plots the empirical distribution and the sample wTO.

Usage

Cut.off(wTO_value, type, plot)

Arguments

wTO_value

is the table returned by the wTO with all the simulated values.

type

type of plot

plot

logical value indicating whether to plot the data.

Value

plots in a new device the cutoff value, and the amount of links in each one of the bands.

ExampleGRF

Description

ExampleGRF data.frame containing data.frame containing names of GRFs.

Usage

data(ExampleGRF)

Format

data.frame 184 lines, 1 column.

Microarray_Expression1

Description

Microarray_Expression1 data.frame containing expression data for 1000 genes and 18 individuals.

Usage

Microarray_Expression1

Format

data.frame 1000 lines, 18 columns.

Microarray_Expression2

Description

Microarray_Expression2 data.frame containing expression data for 1000 genes and 18 individuals.

Usage

Microarray_Expression2

Format

data.frame 1000 lines, 18 columns.

NetVis

Description

Given a set of Nodes and the weight of the edges, a cutoff for the edges, it draws the networks. Returns a list with the nodes and edges attributes. And plots the network.

Usage

NetVis(
  Node.1,
  Node.2,
  wTO,
  pval = NULL,
  MakeGroups = FALSE,
  padj = NULL,
  cutoff = list(kind = "Threshold", value = 0.5),
  layout = NULL,
  smooth.edges = T,
  path = NULL,
  Cluster = F,
  legend = T,
  shape = list(shape = "triangle", names = NULL),
  manipulation = F
)

Arguments

Node.1

Names of the Nodes.1 that are connected to the Nodes.2. It's the output from wTO.Complete or Consensus.

Node.2

Names of the Nodes.2 that are connected to the Nodes.1. It's the output from wTO.Complete or Consensus.

wTO

weight of the links, the wTO output from wTO.Complete or wTO.Consensus.

pval

p-values for the wTO value. By default it is NULL.

MakeGroups

algorithm to find clusters. One of the followings: walktrap, optimal, spinglass, edge.betweenness, fast_greedy, infomap, louvain, label_prop, leading_eigen. Default to FALSE.

padj

Adjusted p-values for the wTO value. By default it is NULL.

cutoff

It's a list containing the kind of cutoff to be used (pval, Threshold or pval.adj)and it's value. Example: cutoff= list(kind = "Threshold", value = 0.5)

layout

a layout from the igraph package.

smooth.edges

If the edges should be smoothed or not.

path

If the graph should be saved specify the name of the file.

Cluster

TRUE or FALSE if the nodes should be clustered (double click to uncluster).

legend

TRUE or FALSE if the legend should appear.

shape

a list shape=list(shape = "triangle", names = NULL), with the shape and the IDs that should have a different shape, shape can be: diamond, star, triangle, triangleDown or square.

manipulation

TRUE or FALSE if the graph should be editable.

Author(s)

Deisy Morselli Gysi <deisy at bioinf.uni-leipzig.de>

Examples

## Not run: 
 X =  wTO.Complete( k =1, n = 5, Data = Microarray_Expression1,
 Overlap = ExampleGRF$x[1:10], method = "p",  plot = FALSE)
# Plot with the default aguments.
 NetVis(Node.1 = X$wTO$Node.1, Node.2 = X$wTO$Node.2,
 wTO = X$wTO$wTO_sign, cutoff = list(kind =
"Threshold", value = 0.50))

# Plotting just the edges with p-value < 0.05, with straight edges, nodes clustered,
# no legend and mapipulation of the graph enabled.
  NetVis(Node.1 = X$wTO$Node.1, Node.2 = X$wTO$Node.2,
 wTO = X$wTO$wTO_sign, pval = X$wTO$pval_sign,
 padj = X$wTO$pval_sign,
  cutoff= list(kind = "pval", value = 0.05),
  smooth.edges = FALSE,
 Cluster = TRUE, legend = FALSE, manipulation = TRUE)
# Plotting just the edges with wTO > 0.50, no legend and the nodes:
# "ZNF738", "ZNF677" with triagle shape,
# no legend and mapipulation of the graph enabled.
 NetVis(Node.1 = X$wTO$Node.1, Node.2 = X$wTO$Node.2,
 wTO = X$wTO$wTO_sign, pval = X$wTO$pval_sign,
 padj = X$wTO$pval_sign, cutoff= list(kind = "Threshold", value = 0.5),legend = FALSE,
 shape = list(shape = "triangle", names = c("ZNF738", "ZNF677")))

 
## End(Not run)

metagenomics_abundance

Description

metagenomics_abundance

Usage

data('metagenomics_abundance')

Format

data.frame from The USC Microbial Observatory. The data is public available at <https://www.ebi.ac.uk/metagenomics/projects/ERP013549>

wTO.aux.each

Description

Computes the repeated measures correlation. based on Blendman Altman (1995). Implemented by <https://doi.org/10.3389/fpsyg.2017.00456>

Usage

rmcor(ID, Measure1, Measure2)

sample_ind

Description

Computes the resufling of the expression values for the IDs.

Usage

sample_ind(x, dfExpression)

Arguments

x

Column to be resampled

dfExpression

data.frame object containing the genes expression on the rows and the individuals (Individuals in the Columns)

Author(s)

Deisy Morselli Gysi <deisy at bioinf.uni-leipzig.de>

wTO

Description

Calculates the weighted topologycal overlap (wTO) between a set of Nodes and the Overlapping nodes. This function implements the method from Nowick (2009).

Usage

wTO(A_TF, sign = c("abs", "sign"))

Arguments

A_TF

Is the weighted adjency matrix (correlation matrix).

sign

("abs", "sign") if the user wants to use the absolute correlation or the signed correlation.

Value

A matrix containing the wTO values.

Author(s)

Deisy Morselli Gysi <deisy at bioinf.uni-leipzig.de>

References

Katja Nowick, Tim Gernat, Eivind Almaas and Lisa Stubbs (2009) <doi:10.1073/pnas.0911376106>

wTO.Complete

Description

Compute the wTO and also the bootstraps. Proposed at: arXiv:1711.04702

Usage

wTO.Complete(
  k = 1,
  n = 100,
  Data,
  Overlap = row.names(Data),
  method = "p",
  method_resampling = "Bootstrap",
  pvalmethod = "BH",
  savecor = F,
  expected.diff = 0.2,
  lag = NULL,
  ID = NULL,
  normalize = F,
  plot = T
)

Arguments

k

Number of threads to be used for computing the weight Topological Overlap. Default is set to 1.

n

Number of resamplings, used to compute the empirical distribuitions of the links. Default is set to 100.

Data

data.frame containing the count / expression data for the correlation.

Overlap

Set of nodes of interest, where the Overlapping weights will be computed.

method

Type of the correlation that should be used. "s" / "spearman" will compute the rank spearman correlation, "p" / "pearson" will compute the linear correlation. If no value is given, the default is to use "p".

method_resampling

method of the resampling. Bootstrap, BlockBootstrap or Reshuffle. Bootstrap null hypothesis is that the wTO is random, and Reshuffle tests if the wTO is equal to zero.

pvalmethod

method to compute the multiple test correction for the pvalue. for more information check the function p.adjust.

savecor

T/F if need to save the correlation.

expected.diff

Difference expected between the real wTO and resampled wTO By default, it is set to 0.2.

lag

time dependency, lag, if you are using the BlockedBootstrap.

ID

ID of the samples for the blocked bootstrap (for repeated measures).

normalize

T/F Should the data be normalized?

plot

T/F Should the diagnosis plot be plotted?

Value

a list with results.

Author(s)

Deisy Morselli Gysi <deisy at bioinf.uni-leipzig.de>

Examples

## Not run: 
# Using spearman rank correlation and bonferroni correction for the pvalues.
wTO.Complete( k =8, n = 1000, Data = Microarray_Expression1,
 Overlap = ExampleGRF$x, method = "s", pvalmethod = "bonferroni")
 # Changing the resampling method to Reshuffle.
wTO.Complete( k =1, n = 1000, Data = Microarray_Expression1,
Overlap = ExampleGRF$x, method_resampling = "Reshuffle")
 # Changing the resampling method to BlockBootstrap, with a lag of 2.
 row.names(metagenomics_abundance) = metagenomics_abundance$OTU
 metagenomics_abundance = metagenomics_abundance[,-1]
wTO.Complete( k =1, n = 1000, Data = metagenomics_abundance, method = "s",
Overlap = row.names(metagenomics_abundance), method_resampling = "BlockBootstrap", lag = 2)
wTO.Complete( k =2, n = 1000, Data = Microarray_Expression1, method = "s",
Overlap = ExampleGRF$x, method_resampling = "BlockBootstrap", ID = rep(1:9,each = 2))
X = wTO.Complete( k =1, n = 1000, Data = Microarray_Expression1,
Overlap = ExampleGRF$x, method = "p", plot = FALSE)
 
## End(Not run)

wTO.Consensus

Description

Consensus requires a list of data.frame containing the pair of nodes, and the wTO values for all networks that need to be joined. Reference: arXiv:1711.04702

Usage

wTO.Consensus(data)

Arguments

data

list of data.frame containing the "Node.1", "Node.2" and "wTO".

Author(s)

Deisy Morselli Gysi <deisy at bioinf.uni-leipzig.de>

Examples

## Not run: 
EXAMPLE =  wTO.Complete( k =1, n = 200, Data = Microarray_Expression1,
                                      Overlap = ExampleGRF$x, method = "p")

# Constructing the consensus network
data = list(data.frame(Node.1 = EXAMPLE$wTO$Node.1,
Node.2 = EXAMPLE$wTO$Node.2,
wto_sig = EXAMPLE$wTO$wTO_sign,
pvalsig = EXAMPLE$wTO$pval_sig),
data.frame(Node.1 = EXAMPLE$wTO$Node.1,
Node.2 = EXAMPLE$wTO$Node.2,
wtoabs = EXAMPLE$wTO$wTO_abs,
pvalabs = EXAMPLE$wTO$pval_abs) )
CONS = wTO.Consensus(data)


## End(Not run)

wTO.aux.each

Description

wTO.aux.each calculate the wTO for each one of the resamplings.

Usage

wTO.aux.each(n, Data, Overlap, method, method_resampling, lag, ID)

Arguments

n

Number of bootstraps / reshuffles to be run for the estimates of the "Threshold" or "pval".

Data

data.frame containing the count / expression data for the correlation.

Overlap

Nodes of interested, where the Overlapping weights will be computed.

method

Type of the correlation that should be used. "s" / "spearman" will compute the rank spearman correlation, "p" / "pearson" will compute the linear correlation. If no value is given, the default is to use "s".

method_resampling

method of the resampling. Bootstrap or Reshuffle. Bootstrap null hypothesis is that the wTO is random, and Reshuffle tests if the wTO is equal to zero.

wTO.export

Description

Exports the significative interactions, the wTO weight and pvalues into a .txt file, tab separeted. This file can be imported in other visualization tools (Cytoscape for example).

Usage

wTO.export(DATA, path, sign = TRUE, pvalue = 0.05, padj = 0.05, prop.NA = 0.5)

Arguments

DATA

Output from the function wTO.Complete or wTO.Consensus.

path

Path and file name where the .txt file should be saved.

sign

Should the network contain the results for the signed network or unsigned? Only for data coming from wTO.Complete.

pvalue

cutoff p-value for the network. Only for data coming from wTO.Complete.

padj

cutoff adjusted p-value for the network. Only for data coming from wTO.Complete.

prop.NA

cutoff proportion of NAs for the network. Only for data coming from wTO.Consensus.

Examples

## Not run: 
EXAMPLE =  wTO.Complete( k =1, n = 200, Data = Microarray_Expression2,
                                      Overlap = ExampleGRF$x, method = "p")
wTO.export(EXAMPLE , './EXAMPLE.txt')

#Selection of only the significative ones for the Consensus
Ex_k1_cor_p_boot_p005_sig = subset(EXAMPLE$wTO,
EXAMPLE$wTO$pval_sig < 0.05,
select = c("Node.1", "Node.2", "wTO_sign"))
Ex_k1_cor_p_boot_p005_abs = subset(EXAMPLE$wTO,
EXAMPLE$wTO$pval_abs < 0.05,
select = c("Node.1", "Node.2", "wTO_abs"))
# Constructing the consensus network
CN = wTO.Consensus(data = list(Ex_k1_cor_p_boot_p005_sig,
Ex_k1_cor_p_boot_p005_abs))
wTO.export(CN, './CN.txt')
### You can store the result on the workspace.
y = wTO.export(CN, './CN.txt')
head(y)

## End(Not run)

wTO.fast

Description

Compute the wTO and also the bootstraps. Proposed at arXiv:1711.04702. This is a quicker version of the wTO.Complete. It doesn't contain diagnose plots nor a parallel version.

Usage

wTO.fast(
  Data,
  Overlap = row.names(Data),
  method = "p",
  sign = "sign",
  delta = 0.2,
  n = 10,
  method_resampling = "Bootstrap",
  lag = NULL,
  ID = NULL
)

Arguments

Data

data.frame containing the count / expression data for the correlation.

Overlap

Set of nodes of interest, where the Overlapping weights will be computed.

method

Type of the correlation that should be used. "s" / "spearman" will compute the rank spearman correlation, "p" / "pearson" will compute the linear correlation. If no value is given, the default is to use "p".

sign

Should the wTO be signed?

delta

expected difference between the real wTO and the bootstraped.

n

Number of resamplings, used to compute the empirical distribuitions of the links. Default is set to 100.

method_resampling

method of the resampling. Bootstrap or BlockBootstrap.If the second is used, please give the lag (time dependency among the data).

lag

Time dependency for the blocked bootstrap (for time series).

ID

ID of the samples for the blocked bootstrap (for repeated measures).

Author(s)

Deisy Morselli Gysi <deisy at bioinf.uni-leipzig.de>

Examples

 wTO.fast(Data = Microarray_Expression1,
 Overlap = ExampleGRF$x, 
 method = "p")

#  For a time series with lag = 4
# wTO.fast(Data = Microarray_Expression1,
# Overlap = ExampleGRF$x, 
# method = "p", 
# method_resampling = 'BlockBootstrap', 
# lag = 4)

# For a study where the individuals were measured multiple times.
# wTO.fast(Data = Microarray_Expression1,
# Overlap = ExampleGRF$x, 
# method = "p", 
# method_resampling = 'BlockBootstrap', 
# ID = rep(1:9, each= 2))

wTO.in.line

Description

Transforms a correlation matrix into the line format.

Transforms a correlation matrix into the line format.

Usage

wTO.in.line(d)

Arguments

d

correlation matrix to be converted into the line format.

Value

the wTO matrix into a data.frame: Node1, Node2 and wTO.

the wTO matrix into a data.frame: Node1, Node2 and wTO.

Author(s)

Deisy Morselli Gysi <deisy at bioinf.uni-leipzig.de>

wTO.rep_measure

Description

Compute the wTO for a repeated measures expermiment and also the bootstraps. Proposed at arXiv:1711.04702. This is a quicker version of the wTO.Complete. It doesn'T contain diagnose plots nor a parallel version.

Usage

wTO.rep_measure(
  Data,
  Overlap = row.names(Data),
  ID,
  sign = "sign",
  delta = 0.2,
  n = 10
)

Arguments

Data

data.frame containing the count / expression data for the correlation.

Overlap

Set of nodes of interest, where the Overlapping weights will be computed.

ID

a vector with the individuals identification

sign

Should the wTO be signed?

delta

expected difference between the real wTO and the bootstraped.

n

Number of resamplings, used to compute the empirical distribuitions of the links. Default is set to 100.

Author(s)

Deisy Morselli Gysi <deisy at bioinf.uni-leipzig.de>

Examples


#wTO.rep_measure(Data = Microarray_Expression1, ID = rep(c(1:9),2), 
#Overlap = ExampleGRF$x)