Looking For Correlation Between Genes Features Using Galaxy Tools

Question

[Mycroft34]

Hello,

I would like to look for association of a binding site (from a ChIP-seq experiment) and some features (for example CpG regions) of nearby genes, and eventually do some statistic to check if the association found is at random or not. I am uising galaxy.

I uploaded the data in bed format; I retrieved CpG regions from UCSC, in the same format. As could be expected, the two datasets contain different number of lines.

Now I am stuck; I looked for correlations, but the program proposed in galaxy work only for a unique dataset. Same difficulty if I want to use graphic presentation.

I also tried to generate a unique dataset of regions common to both features, but as expected, both data become linked and a correlation is meaningless, in that case.

My questions:

1) is it something that is possible to study with galaxy ?

2) how would you proceed ?

Thanks for any help.

Answer 1

[Mycroft34]

I finally found my way into this question:

Here is what I have done:

I fetched the mouse CpG (by the way, my question concerned this species) from UCSC main; then I "fetched the closest non overlapping features" between my uploaded dataset and this CpG dataset.

To get a correlation between position of my site and that of nearby CpG, I calculated the distance for each category to the nearest Transcription start site (TSS), and asked for correlation between these last two calculated columns.

As for graphic representation, I just used the scatterplot with these same two columns.

This answer to my own question is given in case a newby (like I am) will find it helpful, and to allow anyone else to critisize my doing; but in this case, please, explain me what I did wrong.

[KCC]

Have you seen this question How do you calculate if two sets of genomic regions overlap significantly? ? Is it similar to what you want?