What does Human Gene Correlation Analysis do?

Human Gene Correlation Analysis (HGCA) is used for the identification of transcriptionally correlated (coexpressed) genes.

How do I cite Human Gene Correlation Analysis?

Michalopoulos, I., Pavlopoulos, G.A., Malatras, A., Karelas, A., Kostadima, M.A., Schneider, R., and Kossida, S. (2012). Human gene correlation analysis (HGCA): A tool for the identification of transcriptionally coexpressed genes. BMC Res Notes 5, 265.

What data are stored and where do they come from?

The database mainly contains:

• Correlation data for each pair of Probe Sets of Affymetrix Human Genome U133 Plus 2.0 Array Chip calculated from 1959 microarray data sets from GEO
• Annotation data of Affymetrix Human Genome U133 Plus 2.0 Array Chip from Affymetrix
• Genomic data from ENSEMBL
• GO data from Gene Ontology Database
• EC data from KEGG
• InterPro data from EBI
• OMIM data from NCBI
• Transcription Factor Position-Weight Matrix data from TransFac
• Candidate Transcription Factor binding sites of gene regulatory sequences which are found by Match Algorithm

What is the initial search's input and output?

Both tools have an initial search tool that helps the user to identify the probe set of interest and also allows him/her to set the parametres of the preferred analysis.The Probe Set ID (if known), the Gene Name, or any word of the Gene Description (Annotation) can be used as input. The user can set a Summary P-value Cut-off, if a over-representation summary is required. The list tool also inputs the number of most correlated probe sets to the probe set of interest and the tree tool the number of tree nodes between the probe set leaf and the root. The output shows a list of probe set ID, together with their HUGO Gene names and descriptions. When a probe set is clicked, the first analysis is performed.

How are r-, p- and e-values calculated?

Human Gene Correlation Analysis is based on the Pearson Correlation Coefficient (r-value) between the calculated and normalised by MAS 5.0 signal values of the 54613 Probe Sets of Affymetrix Human Genome U133 Plus 2.0 Array Chip from 1959 microarray data sets. The statistical significance of correlation (p-value) is estimated by Student's t-test. Bonferroni correction is used for multiple testing (e-value).

What is the Over-representation Summary and its P-value Cut-off?

If an Over-representation Summary P-value Cut-off is selected, then the tool will perform a term over-representation analysis for each term that is list of the coexpressed genes. The statistical significance (p-value) of the overepresentation of each term is based on Hypergeometric Distibution. The Over-representation Summary only outputs terms whose overrepresentation p-value is below the P-value Cut-off.

Which is the difference between the list and the tree tool?

The list tool produces a list of a preselected number of the most correlated probe sets to a probe set of interest. The tree tool produces a subtree of clustered correlated probe sets which contains the probe set of interest.

What do I get from the colour scheme?

The background of the data from the probe set of interest is pink. The background of alternating tones of green is to distinguish data from each probe set. Dark background is used for the probe sets that one of their genes is in the gene list of another probe set that is already found earlier in the list of coexpressed genes (in list tool) or closer to the tree root (in the tree tool).

What are the Annotation, Biological Process, Cellular Component, Molecular Function, EC Number, OMIM, Pathway, InterPro, TransFac lists useful for?

Nine different sorts of analysis can be performed by each tool:

• Annotation: This is the default analysis of the tools. A description of each gene of each probe set is shown. Overprepresentation analysis shows the most overprepresented words of the collection of descriptions. This analysis is useful when a repeated term can imply an aspect of the nature of the gene of the probe set.
• Biological Process: Biological Process is one of the Gene Ontology aspects. The Biological Process GO Terms for each gene of each probe set is shown. Overprepresentation analysis shows the most overprepresented GO Terms (and their ancestors) of the collection of GO Terms. This analysis is useful when a repeated term can imply the Biological Process the gene of the probe set is likely to participate to.
• Cellular Component: Cellular Component is one of the Gene Ontology aspects. The Cellular Component GO Terms for each gene of each probe set is shown. Overprepresentation analysis shows the most overprepresented GO Terms (and their ancestors) of the collection of GO Terms. This analysis is useful when a repeated term can imply the Cellular Component the gene of the probe set is likely to be part of.
• Molecular Function: Molecular Function is one of the Gene Ontology aspects. The Molecular Function GO Terms for each gene of each probe set is shown. Overprepresentation analysis shows the most overprepresented GO Terms (and their ancestors) of the collection of GO Terms. This analysis is useful when a repeated term can imply the Molecular Function the gene of the probe set is likely to have.
• EC Number: The EC Numbers and their descriptions for each gene of each probe set is shown. Overprepresentation analysis shows the most overprepresented EC Numbers (and their ancestors) of the collection of EC Numbers. This analysis is useful when a repeated term can imply the enzymatic activity of the gene of the probe set is likely to have or the enzyme pathway it participates.
• OMIM: The OMIM Numbers and their descriptions for each gene of each probe set is shown. Overprepresentation analysis shows themost overprepresented OMIM Numbers of the collection of OMIM Numbers. This analysis is useful when a repeated termcan imply a Mendelian Inherited Disease the gene of the probe set is likely to play a role to.
• Pathway: The Pathway terms and their descriptions for each gene of each probe set is shown. Overprepresentation analysis shows the most overprepresented Pathway terms. This analysis is useful when a repeated term can imply the Pathway the probe set is likely to participate to.
• InterPro: The InterPro terms and their descriptions for each gene of each probe set is shown. Overprepresentation analysis shows the most overprepresented InterPro terms (and their offsprings) of the collection of InterPro terms. This analysis is useful when a repeated term can imply the evolutionary origin of the coexpression. Alternatively, it can imply that an apparent correlation between evolutionarily related genes is an artifact, due to cross-hybridisation.
• TransFac: The TransFac Names and their descriptions of the transcription factos that bind to the regulatory sequences of each gene of each probe set is shown. Overprepresentation analysis shows the most overprepresented TransFac Names. This analysis is useful when a repeated term can imply the list of transcription factors which drive coexpression.

How can I navigate through the lists?

Once in the Annotation analysis, the user can change analysis by clicking Annotation, Biological Process, Cellular Component, etc. He/she can also change Summary P-value Cut-off by clicking the appopriate link in the end of the page. Alternatively, it can change probe set, by clicking on its name. The user can visit external sources that are related to the terms shown on the analysis.

How can I navigate through the trees?

Further to the list navigation, the user can choose to see more or less nodes of the subtree. The Newick formatted subtree can also be downloaded.

Which tool should I use?

Both tools are complementary. In practice, the probe sets that have high r-values (>0.6) with their most correlated probe sets, perform better with the tree tool, while probe sets with lower r-values tend to perform better with the list tool.

What are your contact details?

Contact Dr Ioannis Michalopoulos