Bioinformatics Sannio




Synthetic RNA-Seq Network Generation and Mutual Information Estimates

This package implements various estimators of mutual information, such as the maximum likelihood and the Millow-Madow estimator, various Bayesian estimators, the shrinkage estimator, and the Chao-Shen estimator. It also offers wrappers to the kNN and kernel density estimators. Furthermore, it provides various index of performance evaluation such as precision, recall, FPR, F-Score, ROC-PR Curves and so on. Lastly, it provides a brand new way of generatic synthetic RNA-Seq Network with know dependence structure.


either downloaded from CRAN at this link,
The package is implemented in R language and can be either downloaded from CRAN at this link or from GitHub at this link (the *.tar.gz is the compiled package to use with the install package function from R). 


For a quick run of the procedure execute the commands:


simData <- simulatedData(p = 50, n = 100, mu = 100, sigma = 0.25, ppower = 0.73, noise = FALSE)
counts <- simData$counts
adjMat <- simData$adjMat

netData <- mainNetFunction(counts, adjMat, nchips = 25)

plotROC(netData$valMet$valKD, col = "red")

The previous code will generate a network of p=50 nodes (genes) and a count matrix of p=50 genes and n=100 samples. Then it will return the network estimated by all the 10 Mutual Information Methods and evaluate their performances. Lastly, as an example, there is the plot of the ROC Curve of the Kernel Density method.


Garofano Luciano, Pagnotta Stefano Maria and Ceccarelli Michele. “On the Impact of Mutual Information Estimates on Transcriptional Regulatory Network Inference from RNA Sequencing Data” Under Preparation.