| Version: | 1.1.1 |
| Date: | 2024-10-21 |
| Title: | Parallel Mutual Information Estimation for Gene Network Reconstruction |
| Description: | Parallel estimation of the mutual information based on entropy estimates from k-nearest neighbors distances and algorithms for the reconstruction of gene regulatory networks (Sales et al, 2011 <doi:10.1093/bioinformatics/btr274>). |
| License: | AGPL-3 |
| URL: | https://github.com/sales-lab/parmigene |
| BugReports: | https://github.com/sales-lab/parmigene/issues |
| NeedsCompilation: | yes |
| Packaged: | 2024-10-21 13:18:54 UTC; sales |
| Author: | Gabriele Sales [aut, cre], Chiara Romualdi [aut] |
| Maintainer: | Gabriele Sales <gabriele.sales@unipd.it> |
| Repository: | CRAN |
| Date/Publication: | 2024-10-21 13:50:07 UTC |
Algorithm for the Reconstruction of Accurate Cellular Networks
Description
A function that implements the ARACNE algorithm for the reconstruction of gene interaction networks (additive model).
Usage
aracne.a(mi, eps=0.05)
Arguments
mi |
matrix of the mutual information. |
eps |
a positive numeric value used to remove the weakest edge of each triple of nodes. |
Details
This algorithm considers each triple of edges independently and removes the weakest one if
MI(i;j) < MI(j;k) - \varepsilon
and
MI(i;j) < MI(i;k) - \varepsilon
By default, the function uses all the available cores. You can
set the actual number of threads used to N by exporting the
environment variable OMP_NUM_THREADS=N.
Value
A square weighted adjacency matrix of the inferred network.
References
Adam A. Margolin, Ilya Nemenman, Katia Basso, Chris Wiggins, Gustavo Stolovitzky, Riccardo Dalla Favera, and Andrea Califano. Aracne : An algorithm for the reconstruction of gene regulatory networks in a mammalian cellular context. BMC Bioinformatics, 2006.
See Also
Examples
mat <- matrix(rnorm(1000), nrow=10)
mi <- knnmi.all(mat)
grn <- aracne.a(mi, 0.05)
Algorithm for the Reconstruction of Accurate Cellular Networks
Description
A function that implements the ARACNE algorithm for the reconstruction of gene interaction networks (multiplicative model).
Usage
aracne.m(mi, tau=0.15)
Arguments
mi |
matrix of the mutual information. |
tau |
a positive numeric value used to remove the weakest edge of each triple of nodes. |
Details
This algorithm considers each triple of edges independently and removes the weakest one if
MI(i;j) < MI(j;k) * (1 - \tau
and
MI(i;j) < MI(i;k) * (1 - \tau)
By default, the function uses all the available cores. You can
set the actual number of threads used to N by exporting the
environment variable OMP_NUM_THREADS=N.
Value
A square weighted adjacency matrix of the inferred network.
References
Adam A. Margolin, Ilya Nemenman, Katia Basso, Chris Wiggins, Gustavo Stolovitzky, Riccardo Dalla Favera, and Andrea Califano. Aracne : An algorithm for the reconstruction of gene regulatory networks in a mammalian cellular context. BMC Bioinformatics, 2006.
See Also
Examples
mat <- matrix(rnorm(1000), nrow=10)
mi <- knnmi.all(mat)
grn <- aracne.m(mi, 0.15)
Context Likelihood or Relatedness Network
Description
A function that infers the interaction network using the CLR algorithm.
Usage
clr(mi)
Arguments
mi |
matrix of the mutual information. |
Details
CLR computes the score
sqrt(z_i^2 + z_j^2)
for each pair of variables i, j, where
z_i = max(0, ( I(X_i;X_j)-mean(X_i) ) / sd(X_i))
and mean(X_i) and sd(X_i) are the mean
and the standard deviation of the mutual information values
I(X_i;X_k) for all k=1,\ldots,n.
By default, the function uses all the available cores. You can
set the actual number of threads used to N by exporting the
environment variable OMP_NUM_THREADS=N.
Value
A square weighted adjacency matrix of the inferred network.
References
Jeremiah J. Faith, Boris Hayete, Joshua T. Thaden, Ilaria Mogno, Jamey Wierzbowski, Guillaume Cottarel, Simon Kasif, James J. Collins, and Timothy S. Gardner. Large-scale mapping and validation of escherichia coli transcriptional regulation from a compendium of expression profiles. PLoS Biology, 2007.
See Also
Examples
mat <- matrix(rnorm(1000), nrow=10)
mi <- knnmi.all(mat)
grn <- clr(mi)
Parallel Mutual Information Estimation
Description
A function to perform a parallel estimation of the mutual information
of vectors x and y using entropy estimates from K-nearest neighbor
distances.
Usage
knnmi(x, y, k=3, noise=1e-12)
Arguments
x |
a numeric vector. |
y |
a numeric vector with the same length of |
k |
the number of nearest neighbors to be considered to estimate the
mutual information. Must be less than the number of elements of |
noise |
the magnitude of the random noise added to break ties. |
Details
The function adds a small random noise to the data in order to break ties due to limited numerical precision.
By default, the function uses all available cores. You can
set the actual number of threads used to N by exporting the
environment variable OMP_NUM_THREADS=N.
References
Kraskov, Alexander and Stogbauer, Harald and Grassberger, Peter. Estimating mutual information. Phys. Rev. E, 2004.
See Also
Examples
x <- rnorm(100)
y <- rnorm(100)
knnmi(x, y, 5)
Parallel Mutual Information Estimation Between All Matrix Rows
Description
A function that computes the mutual information between all pairs
of rows of matrix mat using entropy estimates from K-nearest neighbor distances.
Usage
knnmi.all(mat, k=3, noise=1e-12)
Arguments
mat |
a numeric matrix (for the reconstruction of gene regulatory networks, genes on rows and samples on columns). |
k |
the number of nearest neighbors to consider to estimate the
mutual information. Must be less than the number of columns of |
noise |
the magnitude of the random noise added to break ties. |
Details
The function adds a small random noise to the data in order to break ties due to limited numerical precision.
By default, the function uses all the available cores. You can
set the actual number of threads used to N by exporting the
environment variable OMP_NUM_THREADS=N.
References
Kraskov, Alexander and Stogbauer, Harald and Grassberger, Peter. Estimating mutual information. Phys. Rev. E, 2004.
See Also
Examples
mat <- matrix(rnorm(1000), nrow=10)
knnmi.all(mat, 5)
Parallel Mutual Information Estimation Between the Rows of Two Matrices
Description
A function that estimates the mutual information between all pairs of
rows of matrices mat1 and mat2 using entropy estimates
from K-nearest neighbor distances.
Usage
knnmi.cross(mat1, mat2, k=3, noise=1e-12)
Arguments
mat1 |
a numeric matrix (for the reconstruction of gene regulatory networks, genes on rows and samples on columns). |
mat2 |
a numeric matrix with the same number of columns as |
k |
the number of nearest neighbors to consider to estimate the
mutual information. Must be less than the number of columns of |
noise |
the magnitude of the random noise added to break ties. |
Details
The function adds a small random noise to the data in order to break ties due to limited numerical precision.
By default, the function uses all the available cores. You can
set the actual number of threads used to N by exporting the
environment variable OMP_NUM_THREADS=N.
References
Kraskov, Alexander and Stogbauer, Harald and Grassberger, Peter. Estimating mutual information. Phys. Rev. E, 2004.
See Also
Examples
mat1 <- matrix(rnorm(1000), nrow=10)
mat2 <- matrix(rnorm(1000), nrow=10)
knnmi.cross(mat1, mat2, 5)
Maximum Relevance Minimum Redundancy
Description
A function that infers the interaction network using the MRNET algorithm.
Usage
mrnet(mi)
Arguments
mi |
matrix of the mutual information. |
Details
The MRNET approach starts by selecting the variable X_i
having the highest mutual information with the target Y.
Then, it repeatedly enlarges the set of selected variables S by
taking the X_k that maximizes
I(X_k;Y) - mean(I(X_k;X_i))
for all X_i already in S.
The procedure stops when the score becomes negative.
By default, the function uses all the available cores. You can
set the actual number of threads used to N by exporting the
environment variable OMP_NUM_THREADS=N.
Value
A square weighted adjacency matrix of the inferred network.
References
H. Peng, F.long and C.Ding. Feature selection based on mutual information: Criteria of max-dependency, max relevance and min redundancy. IEEE transaction on Pattern Analysis and Machine Intelligence, 2005.
See Also
Examples
mat <- matrix(rnorm(1000), nrow=10)
mi <- knnmi.all(mat)
grn <- mrnet(mi)