Softwares:

1. Spectral Analysis for Class Discovery and Classification (SPACC)
2. Fast Calculation of Pairwise Mutual Information Based on Kernel Estimation

1. Spectral Analysis for Class Discovery and Classification (SPACC)

We propose the concept of semi-supervised classification, and build a classifier based-on network connectivity. Similar with supervised learning methods, the proposed method also uses the class label information for training. The major difference is that, in the proposed semi-supervised method, the class label information plays a much less important role compared with supervised methods. By counting the class label information less, the proposed method gained the ability of discovering data subtypes, and better handling data outliers.  

The software can be downloaded at http://www.stanford.edu/~qiupeng/software/SPACC/

2. Fast Calculation of Pairwise Mutual Information Based on Kernel Estimation

We present a new software implementation for more efficiently computing the mutual information for all pairs of genes from gene expression microarrays. Computation of the mutual information is a necessary first step in various information theoretic approaches for reconstructing gene regulatory networks from microarray data. When the mutual information is estimated by kernel methods, computing the pairwise mutual information is quite time-consuming. Our implementation significantly reduces the computation time. The essential idea is simple. In practice, when we encounter computational structures such as nested for-loops, double summations and double integrals, switching the order of the procedures may sometimes result in significant gain. Our fast implementation utilizes this idea to reduce the repeated operations. For an example data set of 9563 genes and 336 samples, the current available software for ARACNE requires 142 hours to compute the mutual information between all possible pairs of genes, whereas our implementation requires 1.6 hours.

The Matlab code is available at http://www.stanford.edu/~qiupeng/software/FastPairMI/