A new network estimation system combined with hierarchical clustering and Graphical Gaussian Modeling , a technique in Multi-variable analysis

　While common statistical analysis seen in gene expression analysis often use correlation coefficient, it is difficult to distinguish the relationship among variables from the one having direct relationship and the one having indirect relationship through other variables. Auto Net Finder is a new tool to distinguish direct and indirect relationship among variables by applying a statistical technique, called Graphical Gaussian Modeling. This software has been developed by the collaboration with Katsuhisa Horimoto, Dr.Sci., National Institute of Advanced Industrial Science and Technology ,Computational Biology Research Center (CBRC)

What is Auto Net Finder???

Estimation in optimal number of clusters by using VIF(Variance of inflation factor)

Visualization among clusters by using correlation coefficient and partial correlation coefficient

Result of PC algorithm

Estimation in optimal number of clusters by the statistical indicator, Variance of Inflation Factor (VIF).
Graphical Gaussian Modeling to estimate the network among variables, a technique in the multi variable-analysis for the next generation.
The interface having many functions to enable high visualization.
Fast and stable calculation, even with the desktop machine.

[Graphical Gaussian Modeling (GGM)]

Graphical Gaussian Modeling efficiently predicts the correlation structure behind the correlation matrix by using the covariance-selection, which enables it to demonstrate the result as a graph subsequently. (Covariance selection is the procedure to select the model of correlation structure, to which has been substituted the partial correlation coefficient be 0 from the observed data.) While the correlation coefficient is the statistical measure to quantify the relationship between two variables, it may give the falsely-positive relationships between two variables, influenced by other variables. Partial correlation enables it to remove this falsely-positive correlation and establish the actual correlation relationship among variables.

[PC algorithm]

Auto Net Finder version2.0 can calculate Graphical Gaussian Model and causation graph by PC algorithm. For example, Auto Net Finder version2.0 can predict the regulation interaction between regulators and target proteins from gene expression data(DNA array)

[[Reference literature]]: [1] Horimoto, K. and Toh, H., Statistical estimation of cluster boundaries in gene expression profile data, Bioinformatics, 17:1143‐1151, 2001.; [2] Horimoto, K., Aburatani, S., Kuhara, S., and Toh, H., ASIAN - automatic system for inferring a network from gene expression profiles, Res. Commun. Biochem. Cell Mol. Biol., 5:192‐207, 2001.; [3] Toh, H. and Horimoto K., System for automatically inferring a genetic network from expression profiles, J. Biol. Phys., 28:449‐464, 2002.; [4] Toh, H. and Horimoto, K., Inference of a genetic network by a combined approach of cluster analysis and graphical gaussian modeling, Bioinformatics, 18:287‐297, 2002.; [5] Aburatani, S., Kuhara, S., Toh, H., and Horimoto, K., Deduction of a gene regulatory relationship framework from gene expression data by the application of graphical Gaussian modeling, Signal Processing, 83:777‐788, 2003.; [6] Sachiyo Aburatani, Fuyan Sun, Shigeru Saito, Masao Honda, Shu-ichi Kaneko, and Katsuhisa Horimoto,, Gene systems network inferred from expression profiles in hepatocellular carcinogenesis by graphical Gaussian model
EURASIP Journal on Bioinformatics and Systems Biology, 2007
[[Operation environment]]: CPU: 500MHz or above; RAM: 256MB or above OS:Windows2000/XP, Linux