# Functions
## panda_run
Description:
Using PANDA to infer gene regulatory network. 1. Reading in input data (expression data, motif prior, TF PPI data) 2. Computing coexpression network 3. Normalizing networks 4. Running PANDA algorithm 5. Writing out PANDA network (optional)
Inputs:
exp_file : path to file containing gene expression as a matrix of size (g,g) motif_file: path to file containing the prior TF-gene regulatory network based on TF motifs as a matrix of size (t,g) ppi_file : path to file containing TF-TF interaction graph as a matrix of size (t,t) panda_out : path to save output PANDA network
‘.txt’: the final network will be saved in .txt format ‘.tsv’: the final network will be saved in .tsv format ‘.’ : the final network will be saved in .mat v6 format ‘’ : the final network will not be saved
- save_temppath to save updated ppi, co-expression, and gene regulation network
‘’: the networks will not be saved
alpha : learning parameter for the PANDA algorithm save_pairs: (Optional) boolean parameter
1: the final network will be saved .pairs format where each line has a TF-gene edge (Cytoscape compatible) 0: the final network will not be saved in .pairs format
Outputs:
AgNet : Predicted TF-gene gene complete regulatory network using PANDA as a matrix of size (t,g).
Authors:
cychen, marieke, kglass
Notes:
Script adapted from Marieke’s pLIONESSpcss.m, modified to run PANDA only.
Publications:
## lioness_run
Description:
Using LIONESS to infer single-sample gene regulatory networks. 1. Reading in PANDA network and preprocessed middle data 2. Computing coexpression network 3. Normalizing coexpression network 4. Running PANDA algorithm 5. Writing out LIONESS networks
Inputs:
exp_file : path to file containing gene expression as a matrix of size (g,g) motif_file: path to file containing the prior TF-gene regulatory network based on TF motifs as a matrix of size (t,g) ppi_file : path to file containing TF-TF interaction graph as a matrix of size (t,t) panda_file: path to the PANDA generated gene regulatory network save_dir : path to save directory. if It does not exist, it will be created. START : index of first sample to generate predicted gene regulatory network. END : index of last sample to generate predicted gene regulatory network. There will be END-START+1 single network samples generated.
-1: use the index of the final gene expression sample
alpha : learning parameter for the PANDA algorithm ascii_out : 1 : save LIONESS networks in .txt format
0 : save LIONESS networks in .mat -v6 format
lib_path : path to library
Outputs:
- PredNetPredicted single sample network as a matrix of size (t,g)
This output is directly saved as a file and not as worksapce variable
Authors:
cychen, marieke, kglass
Publications:
## RunPUMA
Description:
PUMA can reconstruct gene regulatory networks using both transcription factors and microRNAs as regulators of mRNA expression levels.
Inputs:
exp_file : path to file containing gene expression as a matrix of size (g,g) motif_file: path to file containing the prior TF-gene regulatory network based on TF motifs as a matrix of size (t,g) ppi_file : path to file containing TF-TF interaction graph as a matrix of size (t,t) outtag : path to save output PUMA network in .pairs format. mir_file : path to file containing microRNA file alpha : learning parameter for the PUMA algorithm
Outputs:
RegNet : Predicted TF-gene gene complete regulatory network using PANDA as a matrix of size (t,g).
Authors:
Marieke Kuijjer
Publications:
## UpdateDiagonal
Description:
Updates the diagonal of diagMat
Inputs:
diagMat: diagonal matrix num : integer alpha : learning rate of the PANDA algorithm step : step number in the PANDA algorithm
Outputs:
diagMat: updated diagonal matrix
Authors:
Kimberley Glass
## Tfunction
Description:
Updates Amat matrix using matrices X and Y
Inputs:
X : adjacency matrix Y : adjacency matrix
Outputs:
Amat: updated matrix from X and Y
Authors:
Kimberly Glass, cychen
Notes:
bsxfun is more memory-efficient and faster than repmat implementation for large arrays. MATLAB uses BLAS routines to do matrix multiplication. MATLAB parser recognizes X*X’ as a symmetric matrix multiply and will call the symmetric matrix multiply routine (only calculates about 1/2 the answer and then fills in the rest with copies, which is faster).
## SavePairs
Description:
A function to save a complete graph in matrix format to a pairs format where each line represents an edge in the network. The output file will have as much lines as edges in the network and will have the predicted edge weights as well as the binary edge weights from the prior motif data.
Inputs:
TFNames : names of t TFs GeneNames: names of g genes AgNet : predicted gene regulation network using PANDA of size (t,g) RegNet : prior gene regulation network obtained using TF motif scan of size (t,g) outtag : name of saved file
Authors:
Kimberley Glass, Marouen Ben Guebila
## Pairs2Mat
Description:
Pairs2Mat transforms a TF-Gene network in .pairs network to a complete matrix. It can also save the prior in matrix format.
Inputs:
- networkPair: path to network in .pairs format with nGenes*nTFs
edges
nGenes : number of genes in network prior : 0: matNet is a matrix of the final network
1: matNet is a matrix of the prior network
Outputs:
matNet : network in nGenes by nTfs matrix format
Authors:
Marouen Ben Guebila 6/19
## PANDA
Description:
PANDA infers a gene regulatory network from gene expression data, motif prior, and PPI between transcription factors
Inputs:
RegNet : motif prior of gene-TF regulatory network GeneCoReg: gene-gene co-regulatory network TFCoop : PPI binding between transcription factors
Outputs:
RegNet : inferred gene-TF regulatory network
Authors:
Kimberley Glass
Publications:
## PUMA
Description:
PUMA can reconstruct gene regulatory networks using both transcription factors and microRNAs as regulators of mRNA expression levels.
Inputs:
RegNet : motif prior of gene-TF regulatory network GeneCoReg: gene-gene co-regulatory network TFCoop : PPI binding between transcription factors alpha : learning rate s1, s2, t1, t2 are indices of miR interactions in TFCoop, the TF-TF PPI network.
Outputs:
RegNet : inferred gene-TF regulatory network
Authors:
Marieke Kuijjer
Publications:
## NormalizeNetwork
Description:
Normalize an input network (matrix) X
Inputs:
X: network adjacency matrix
Outputs:
normMat: nromalized adjacency matrix
Authors:
Kimberley Glass
## Coexpression
Description:
Compute gene-gene coexpression network for a sample-by-gene matrix X Note that each gene is a column in X
Inputs:
X: sample-by-gene matrix
Outputs:
GeneCoReg: gene-gene coexpression network
Authors:
Kimberley Glass