A Covariance Structure Analysis model builds a pathway relating TCA cycle.

The path model estimated from gene expression data. Expression data are from Eisen et al. with 79 measures. After inputting the gene expression matrix into a covariance structure analysis software EQS, we obtain a statistical acceptable path model based on the presumed pathways from DeRiSi. The paths are consistent with the metabolic pathways. Moreover the path coefficients over the arrows indicate the effect of either induction or repression and the strength of the gene association.

A factor analysis model gives two factors representing 81 cell cycle regulated genes

The array data set includes 81 cell cycle genes from Spellman et al. 1998 in 42 measures of -arrested and Cdc15-based synchronizations. Notations “1”-“6” respectively denote the cycle phases of G1 MCB, G1 SCB, S, S/G2, G2/M, M/G1. The genes are positioned in a circle corresponding to the cell cycle.

Path diagram for 5 reference genes of yeast cell cycle

The 5 genes are from Spellman et al. with 79 expression measures. The arrows indicate the causal relationships of the genes based on their expression measures. The numbers over the arrows indicate the relative strength

Part of the yeast cell cycle pathway constructed by covariance structure model

The sets of highly related genes are indicators of their common factors. Factor 1 is interpreted as the transcript factor MBF, because it represents the group of G1 MCB genes. The path coefficients over the arrow indicate the strength of the association between groups of genes. The sign of the coefficients indicate activating or inhibiting between groups of genes.