Home - About - Screenshots - Tutorial - FAQ - Mailing list - What's new

The query:

Goto yMGV homepage and follow the "Access 4" link (arrow 1 in next picture)


Enter the ORF or gene name you want to use as a reference to find co-expressed genes.
Select the set of experiments you want to search for co-expressed genes. The list of all publications available in each experiment set is available on-line.
Select the distance calculation method you want. Details on distance calculation are here.
Filter output for most variants expression profiles. See the MiCoViTo tutorial for more information on the filtering process.
Apply pre-treatment (centering and scaling) to the data. For more information on the centering and filtering process see the MiCoViTo tutorial.
To filter the list of output genes, choose the distance values you want as threeshold. Details on distance calculation is available here.
You can display the cluster view of all ratio for the co-expressed genes found in the selected set of experiments.
GO informations can also be displayed.

The result:

Information about the selected gene extracted from the SGD, and gene ontology terms associated. Some statistics about variation of its expression accross all the experiments in yMGV database. See "yMGV - FAQ" for more informations.
The distance distribution is displayed as a histogram. The result is of course more accurate for a gene having a strong expression variation compare to a "flat profile" gene. In addition, for larger experiment datasets, result is more accurate.
Table showing the distance value range needed to comply to the statistical alpha value level (0.1, 1 or 5%). More Informations
As for other yMGV modules, gene list can be posted to other external ressources.
Genes found sorted by distance value.
Color coded check boxes indicate the level of confidence for this matched gene according to the statistical test. More Informations

A cluster view of the co-expressed genes in the experiment set is available:

Log2(ratio) color scale.
List of all the experiments included into the selected experiment set.
One row for each of the co-expressed genes found.
One column for each of the experiment of the experiment set choosen.

Correlation coefficient:

Testing the significance of the coefficient correlation:

How to be sure that a significant correlation coefficient indicates the existence of a REAL relationship between two genes ?

This question can be answered by using a modified t-test :

H0 hypothesis : The correlation measure is a sampling error
H1 hypothesis : The correlation measure representes a REAL association

Under the H0 hypothesis, we know that ( r = value of the correlation coefficient , N = number of publications in experiment set ) :

follows the probability distribution of the "Student" t statistic, with N - 2 degrees of freedom.

If the t value measured exceeds the critical value (according to the Student's t distribution), then the hypothesis H0 can be rejected at the specified level of significance (alpha = 5%, 1% or 0.1%). This level of significance is the maximum probability we are willing to risk in deciding to reject the H0 hypothesis (no correlation) when it is true.

Distance calculation method:

• Pearson distance formula : 1 – (Pearson correlation coefficient)


Pearson correlation coefficient


• Squared Pearson distance formula : 1 – (Squared Pearson correlation coefficient)


Squared Pearson correlation coefficient


• Euclidian distance formula :


Euclidian distance