Question on Z score mRNA expression for patient dataset
291 views
Skip to first unread message
Teoh, Jian Peng
Dec 30, 2013, 5:51:17 PM12/30/13
to cbiop...@googlegroups.com
Hello,
My name is Jian Peng Teoh, currently a biomedical PhD student in Georgia Regents University, who is interested in the research of ovarian cancer.
The question that I would like to ask is that I was wondering about the default value of 2 for z-score threshold in regards to the mRNA data expression for Ovarian Serous Cystadenocarcinoma (TCGA, Nature 2011) when i was keying in the query to look at the patent dataset.
Is this value of +/- 2 are indication of the significance for the mRNA expression? if not, what is the best z-score threshold value for this particular mRNA expression dataset in order for it to be significant?
Does that mean that a value of 2 or more in the patient dataset means that particular gene is significantly overexpressed?
Does that also mean that values between 2 and -2 are not significantly different?
Thank you very much if you can get back to my question. I hope you can help to clear my confusion. Thanks again for your time.
Sincerely,
Jian Peng Teoh,
Biomedical Science PhD student
Vascular Biology Center
Georgia Regents University
PH: 706-721-5999
My name is Jian Peng Teoh, currently a biomedical PhD student in Georgia Regents University, who is interested in the research of ovarian cancer.
The question that I would like to ask is that I was wondering about the default value of 2 for z-score threshold in regards to the mRNA data expression for Ovarian Serous Cystadenocarcinoma (TCGA, Nature 2011) when i was keying in the query to look at the patent dataset.
Is this value of +/- 2 are indication of the significance for the mRNA expression? if not, what is the best z-score threshold value for this particular mRNA expression dataset in order for it to be significant?
Does that mean that a value of 2 or more in the patient dataset means that particular gene is significantly overexpressed?
Does that also mean that values between 2 and -2 are not significantly different?
Thank you very much if you can get back to my question. I hope you can help to clear my confusion. Thanks again for your time.
Sincerely,
Jian Peng Teoh,
Biomedical Science PhD student
Vascular Biology Center
Georgia Regents University
PH: 706-721-5999
Nikolaus Schultz
Dec 31, 2013, 1:05:25 PM12/31/13
to cbiop...@googlegroups.com, JT...@gru.edu
Hi Jian,
The definition of "significance" is somewhat arbitrary. The z-score of 2 defines a difference of two standard deviations from the mean of expression of all samples that are diploid for the gene of interest. You can define a z-score of 3 to get an "even more significant" difference. That said, a lot of people use 2 z-scores as a threshold for significance.
Niki.
> --
> 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/groups/opt_out.
The definition of "significance" is somewhat arbitrary. The z-score of 2 defines a difference of two standard deviations from the mean of expression of all samples that are diploid for the gene of interest. You can define a z-score of 3 to get an "even more significant" difference. That said, a lot of people use 2 z-scores as a threshold for significance.
Niki.
> 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/groups/opt_out.
Reply all
Reply to author
Forward
0 new messages