Value to define methylation in cBioportal

[duzhenf...@gmail.com]

Hi, I have extracted methylation data about a specific gene from cBio and I get a value in each sample. I am not sure what this value mean, the reference said each beta-value stands for the methylation status in each locus, while the website said "For genes with multiple probes, the probe most anti-correlated with expression". What's that mean? Can I dichotomize the data using a β-value of >0.3 as a threshold for positive methylated gene?

Thanks very much for your time. Looking forward to your reply.

Sincerely,

Zhenfang

The Chinese University of Hong Kong

[Nikolaus Schultz]

A given gene may have multiple CpG sites tested in its promoter region or even gene body. The results from these probes are beta-values, and they indicate the degree to which each CpG site is methylated (0 = no methylation, 1 = fully methylated). Since cBioPortal only supports one value per gene, we can only display one beta-value. To make sure that we show the most meaningful one, we test the correlation between beta values and mRNA expression values for all probes for a given gene and then pick the one with the most negative correlation. The idea here is that gene silencing by methylation should be visible in the mRNA profile for that gene, and that samples with increased methylation should have lower expression. 

For most genes, we see little correlation, but there are well known examples where beta-values are correlated strongly with expression. See for example MLH1 in colorectal or endometrial cancer, or BRCA1 in ovarian or breast cancer.

So it is probably best to include mRNA expression levels when trying to make a statement about silencing of a gene by methylation. I hope this is useful.

Niki.

P.S.: When responding, please cc the Google Group.

--
You received this message because you are subscribed to the Google Groups "cBioPortal for Cancer Genomics Discussion Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cbioportal+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[duzhenf...@gmail.com]

Dear Niki,

Thanks for your reply. We selected a gene of interest and found that the 12/14 probes located in the promoter region, and also as you said cBio choose the probe with the most negative correlation with mRNA expression. So in this scenario,  think it make sense to use the only one beta-value to define methylation of this gene. Is that OK?

Then there is another question. How to define a gene which is downregulated or upregulated in cBio? I extracted mRNA Expression z-Scores (RNA Seq V2 RSEM) from cBio, It's hard for me to understand z-Scores. From the data, I see z-Score >0 or <0, can I conclude that >0 stands for "upregulated" and <0 stands for "downregulated"?

Sincerely,

Zhenfang

The Chinese University of Hong Kong

Nikolaus Schultz於 2016年7月15日星期五 UTC+8下午9時41分56秒寫道：

[Nikolaus Schultz]

Sorry for the late response.

Yes, positive z-scores indicate higher expression relative to the other tumor samples (not to normal samples). A z-score indicates the number of standard deviations of the mean of the average expression in the reference population. For TCGA samples, cBioPortal uses the expression distribution in all samples that are diploid for a given gene.

Niki.

[Zhenfang Du]

Hi Niki,

Thanks so much for your reply. I'm only focus on promoter CpG methylation. In each gene, cBio shows the probe of anti-correlation with mRNA expression, and HM450K has about 10 probes in each gene, I'm wondering how can we confirm which probe cBio choose to exhibit beta-value?

Sincerely,

Zhenfang

[CB]

Hi Niki, 

Regarding the beta value for methylation of individual CpG sites vs corresponding mRNA expression. This makes sense if the CpG site is within the promoter region since promoter methylation shows an inverse correlation with mRNA expression. However, gene body methylation shows a direct correlation with gene expression. Have you guys considered considered differentiating promoter and gene body methylation? Or possibly just analyzing promoter methylation? 

Thanks, 
Carter

[Nikolaus Schultz]

Good point. For the moment, we are only considering silencing via promoter methylation. We will need to change our data models to accommodate multiple probes per gene, and that is currently not a priority. 

Niki. 

To post to this group, send email to cbiop...@googlegroups.com.
Visit this group at https://groups.google.com/group/cbioportal.

[xavie...@gmail.com]

Dear Niki,

this is an old post, but I was curious to see whether you could clarify on what you precisely consider "promoter" in this statement on the previous email: "we are only considering silencing via promoter methylation". In this case does promoter mean TSS200 (i.e., 200 bp upstream the TSS), TSS1500 (i.e., 1500 bp upstream the TSS), or another combination including regions such as 5'UTR, 1st Exon, etc?

Thanks for your help.

Best,

El divendres, 5 agost de 2016 23:47:33 UTC+2, Nikolaus Schultz va escriure:

[Nikolaus Schultz]

We are currently not assessing where in the gene the probe is - we actually look at all probes available for a given gene and use the one with the strongest negative correlation with expression.

We are considering expanding the methylation data into several categories: promoter, gene body, and possible enhancer… whatever makes sense. But we are still exploring this and any input it welcome.

Niki.

To view this discussion on the web visit https://groups.google.com/d/msgid/cbioportal/0efdf395-7650-4653-978f-3bcffa657063%40googlegroups.com.