Pearson correlation

[Oladipo Oluwakemi]

Dear Sir,

Good day. I will like bring to your attention that the pearson correlation column is missing on the co-expression gene list. I have attached a screenshot to this email. 

I look forward to your kind response as I need this data for my analysis.

--

Thank you. 

Best regards,

Hannah Oluwakemi Oladipo

Senior Scientific Officer,

National Biotechnology Development Agency(NABDA),

Airport road,Lugbe

Abuja,Nigeria

+2348033319207

[JJ Gao]

Hi Hannah,

Thanks for contacting us. We decided to remove Pearson's correlation scores since the RNAseq expression data we have are oftentimes non-linear. Would Spearman's correlation suffice for your analysis? We also plan to add p/q-values for Spearman's correlation soon. Please let us know if Pearson's correlation is necessary for your analysis.

Thanks,

-JJ

--
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.
To post to this group, send email to cbiop...@googlegroups.com.
Visit this group at https://groups.google.com/group/cbioportal.
To view this discussion on the web visit https://groups.google.com/d/msgid/cbioportal/CAG8aZ%3D3E%2BsGR9mML3j1PdiVOQ1xdmzZ4P6e0YD7bRebuXtiH5g%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

[JJ Gao]

Thanks. We plan to add p-values earlier next week. Please stay tuned.

-JJ

On Fri, Sep 14, 2018 at 10:03 AM Oladipo Oluwakemi <kemio...@gmail.com> wrote:

Dear JJ,

Thank you for the kind information. The main focus for me is to be able to select significantly co-expressed genes. So I can work with either Spearman or Pearson correlation but having the P-value of the correlation will be very helpful. How soon are you planing to input the P-values for the Spearman's correlation?

Thank you.

Best regards,

Kemi

[Oladipo Oluwakemi]

Dear JJ,

Thank you for the kind information. The main focus for me is to be able to select significantly co-expressed genes. So I can work with either Spearman or Pearson correlation but having the P-value of the correlation will be very helpful. How soon are you planing to input the P-values for the Spearman's correlation?

Thank you.

Best regards,

Kemi

On Fri, Sep 14, 2018 at 3:10 PM JJ Gao <jianji...@gmail.com> wrote:

[Oladipo Oluwakemi]

Hello,

Good day. I have an observation about the discrepancy between the co-expressed genes obtained from Cbioportal and the correlated genes obtained from GEPIA. The Spearman and Pearson correlation of the co-expressed genes does not correspond to that of GEPIA. Do you have an idea why this could be so? I will like to know this because of my data analysis and to avoid errors in my analysis. I look forward to your kind response.

[JJ Gao]

Hi Kemi,

There could be various reasons, e.g. different versions of TCGA data, different sets of sampels, etc.  Would you please give us an example? A query url of cBioPortal would be useful. Please also let us know the steps of your analysis on GEPIA.

Thanks,

-JJ

[Oladipo Oluwakemi]

Hello,

Thank you for your prompt response. I have attached  to this email a screenshot of the list of the co-expressed genes with TRAP1 (one of the genes I am analyzing) in SKCM cancer as obtained from Cbioportal. Looking at the first gene in this list (TUFM), I have also attached the result of the correlation of the expression of this two genes in GEPIA. You will observe that the Spearman correlation is different in both cases and so also is the p value. In GEPIA, I used only the tumor sample. (NB: there is a correlation tab on GEPIA , that is how I obtained the screenshot) Thank you.

[Nikolaus Schultz]

Hi Oladipo,

It looks like GEPIA uses TPM, while cBioPortal uses RSEM. That probably explains the differences. 

It looks like you are at least getting the same trends, correct?

Niki.

To view this discussion on the web visit https://groups.google.com/d/msgid/cbioportal/CAG8aZ%3D3HqgC8j46qg1CFByo-iejHWOrUXbN6QQzvw7%3D-4X6Lhg%40mail.gmail.com.

For more options, visit https://groups.google.com/d/optout.

<SKCM_TRAP1_TUFM.png><SKCM_TRAP1_TUFM_GEPIA.png>