How do I decide where the cut-off is between significantly correlated genes versus irrelevant genes?
Some genes lower down a correlation list may be regulated by multiple, potentially antagonistic, mechanisms but may still have some regulatory signal in common with the genes at the top of the list. The signal can be a promoter motif common to a set of genes and it can be fruitful to look for any motifs present in promoters of genes at the top and progressively lower down the list. Alternatively, the 2-D co-correlation plot can be used as a visual tool to suggest an r-value which may be used as a cut-off. Use two genes strongly-implicated in the same process as the drivers in the co-correlation plot. Often there will be a cluster or “tail” of other genes strongly-correlated with both drivers and separated from the bulk of the other genes. This may be taken as an empirical cut-off threshold giving you a set of genes to be further analysed by other methods. There is an example on our web-site using two heat-shock genes (http://www.arabidopsis.leeds.ac.uk/act/coexpanalyser.php#CO3, heat s
Related Questions
- How does the doctor decide which treatment is the best, such as whether to use a Pixel laser treatment versus Perlane for an under the eye treatment?
- Im confused by correlation versus causation. If two things are correlated, cant they also have a cause-and-effect relationship?
- Why did Borders decide to launch its own site versus continuing the relationship with Amazon.com?
