Visualizing mutations burden vs. gene expression

Question

[H.Hasani]

Dear biostars,

I know that visualizing SNPs has been asked several times and in different context, but I would like explicitly to combine mutations burden and gene expression per individual. Before I start making my own script, I was wondering, if there is already a tool for that?

Edit: thank you very much for the answers. My purpose is to visualize gene expression and mutations burden, NOT mutations effect OR their association with genes. It is really really simple, I wanted to know if this type of visualization was already done to avoid re-inventing the wheel.

Thanks!

[Denise CS]

As part of the Open Targets project, we've linked different types of evidence (e.g. somatic and germline mutations, RNA expression, animal models, affected pathways, text mining, etc) to a given gene in a given disease e.g. PTEN in neoplasm. Maybe that could be a useful dataset to look at? The data can be retrieved programmatically (check our API documentation) if you are interested.

[H.Hasani]

Thank you very much! I liked it already, it is very helpful, since I can see my genes there but from different prospective.

Thank you for sharing it

[ivivek_ngs]

Ah yes I forgot to mention about Open Targets project. Thanks for putting it up in the thread.

[GouthamAtla]

This may not be an answer to your question but might give some idea about plotting.

How to create a mutation landscape (waterfall) plot with GenVisR

If you replace the mutation type with the expression values, you might achieve what you are looking for.

[H.Hasani]

Thanks! I'm using it, and the package's developer is such a great support.

[ivivek_ngs]

Yes the lolliplots are a good idea. It should serve as a tool. I was more trying to give the relevance of the biological context. Just that ciobioportal and Oncoprints are already available online to map with public datasets and the rest for individual datasets OP can try with the GenVisR. Thanks for bringing it up for the OP.

[H.Hasani]

Exactly, I just got the green light from the package's developer to do so

Answer 1

[poisonAlien]

Hi, take a look at oncoprint function from ComplexHeatmap. It has a tutorial on making a plot combining mutation load and gene expression.

Answer 2

[ivivek_ngs]

There are things which I would like to say here and then you must see to what extent these mutations at each individual is affecting the gene expression:

1. The type of mutation be it a stop loss or a stop gain will affect directly the gene expression be it up or down.
2. If you have a oncogene or tumor suppressor gene then you will also be able to assess the same gene's expression in your individual sample and see how it is changing in different samples.
3. If you have indels that will also directly impact the transcriptional read out.

As far as checking them, I know stuffs related to tumor as I work on them mostly. If they are of your help you can take a look. I am not sure what kind of disease you are looking. You can directly take your genes of interest or transcript id and scan through your expression matrix for row.names (genes/transcripts) to find the hits cross your samples and plot the distribution of bar plots to see the transcriptional readouts or if it does not convey much from your own sample then try to such in public data sets from below links.

Tumor Portal

OASIS from Pfizer

CbioPortal from MSKCC

All the above lets you see gene level expression across in larger patients.

So I believe even if you do not find much in your data due to less samples or whatever constraints , you can still try for public data to prove your hypothesis.

[H.Hasani]

Thanks! I think you are reading what is behind my lines :) I edited the question to avoid confusion.

[ivivek_ngs]

Sorry for over reading the motivation. As for the things like open target one can still take a look at the 3 links I posted if you want to work with public datasets , there are multiple plots and tumorportal.org is the paper which showed the mutational burden across 21 types of cancer so you might find it handy and also get some tools that they used. But yes Goutham's suggestion is very apt then.