BACK  

• Basic Statistics includes raw data of intensity of both signal and backgrounds. The calls for presence of gene, or for changed expression of a comparison, and related p-value of the t-test will be provided. 
• Hierarchical clustering produces a familiar tree structure. This can be done through one of three major similarity metrics plus one of three major linkage methods. It is a way to group genes based on their similarities.
• K-mean/median clustering, a non-hierarchical cluster analysis, partitions the data into mutually exclusive and exhaustive groups (ten or so). Each cluster is approximated by either means or medians.  
• Gene-shaving clustering, a non-hierarchical cluster analysis, differ from other methods by identifying groups with large variation, assuming that clusters with large variance may likely be differentially expressed. One gene may belong to many clusters.  
• PCA is used to produce low dimensional clusters. It generates derived variables (components) that are linear combinations of expression data. No premature categorization of data.
• SOM analysis, an amendment to K-means methods, considers the expression levels of n probe sets in k experiments as n points in k-dimensional space/grid. It is one way of identifying gene expression patterns.
• Matrix analysis computes the probability (p-value) that the observed overlap is expected due to random chance. The matrix provides a spreadsheet framework for comparing probe lists and displays the overlap or non-overlap significance score for probe lists in the matrix. It displays significance threshold (pink) or the non-overlap significance threshold (yellow). 
• SAM analysis computes a statistic di for each gene i, measuring the strength of the relationship between gene expression and the response/treatment variable. The cutoff for significance is determined by a tuning parameter delta, chosen by the user based on the false positive rate. One can also choose a fold change parameter, to ensure that called genes change at least a pre-specified amount.
• Visualization (histogram/scattered/line graph/3D).
• NetAffx of Affymetrix for functional annotations: Annotation table/ Molecular functions / Cellular components / Biological processes /Pathways / Public accession numbers / protein accession numbers. 
• Customer query and results: as investigator desire.
• Newly developed or investigator-identified analysis by Biostatistics and Bioinformatics Unit of the CCC will be assisted.
• Note:  Not all of the above analyses are available for data from both Affymetrix GeneChip and Spotted arrays.  It is suggested that investigators present a desired format of reports, for example, table, graph, image, et al., before choosing these analyses.  Please contact Dr. Dongquan Chen at 934-6842 for further information (dongquan@uab.edu)