Hi All
Is there a way to determine expression status of a gene (up, down)
from
corrected p.values ?
Thanks a lot
Alyaa
--
Alyaa Mahmoud
"Love all, trust a few, do wrong to none"- Shakespeare
[[alternative HTML version deleted]]
Hi Alyaa,
On 7/11/2012 2:59 AM, Alyaa Mahmoud wrote:
> Hi All
>
> Is there a way to determine expression status of a gene (up, down)
from
> corrected p.values ?
No. The p-values are strictly between 0 and 1, so there is no way to
discern the sign of the underlying t-statistic. This would not be true
if you were doing a one-sided t-test, but I can't imagine a use case
for
a one-sided t-test in the context of microarray analysis.
Best,
Jim
>
> Thanks a lot
> Alyaa
>
--
James W. MacDonald, M.S.
Biostatistician
University of Washington
Environmental and Occupational Health Sciences
4225 Roosevelt Way NE, # 100
Seattle WA 98105-6099
I can easily think of a reason for a one-sided test. Consider the
case
where I have copy number data and gene expression data on a bunch of
samples, and I'd like to test whether changes in gene expression go
the
same direction as changes in copy number.
On 7/11/2012 8:21 AM, James W. MacDonald wrote:
> Hi Alyaa,
>
> On 7/11/2012 2:59 AM, Alyaa Mahmoud wrote:
>> Hi All
>>
>> Is there a way to determine expression status of a gene (up, down)
from
>> corrected p.values ?
>
> No. The p-values are strictly between 0 and 1, so there is no way to
> discern the sign of the underlying t-statistic. This would not be
true
> if you were doing a one-sided t-test, but I can't imagine a use case
> for a one-sided t-test in the context of microarray analysis.
>
> Best,
>
> Jim
>
>
>>
>> Thanks a lot
>> Alyaa
>>
>
An embedded and charset-unspecified text was scrubbed...
Name: not available
URL: <https: stat.ethz.ch="" pipermail="" bioconductor="" attachments="" 20120712="" 573f1a31="" attachment.ksh="">
It depends. In CLL, almost all of the copy number changes are single
copy gains or single copy losses (for different genes). So the copy
number data only has two levels, in which case correlation doesn't
tell
you very much and a one-sided t-test is better.
Kevin
On 7/12/2012 12:18 AM, Alyaa Mahmoud wrote:
> wouldn't we try a correlation test better then ?
>
> On Wed, Jul 11, 2012 at 7:10 PM, Kevin R. Coombes
> <kevin.r.coombes@gmail.com <mailto:kevin.r.coombes@gmail.com="">>
wrote:
>
> I can easily think of a reason for a one-sided test. Consider
the
> case where I have copy number data and gene expression data on a
> bunch of samples, and I'd like to test whether changes in gene
> expression go the same direction as changes in copy number.
>
> On 7/11/2012 8:21 AM, James W. MacDonald wrote:
>
> Hi Alyaa,
>
> On 7/11/2012 2:59 AM, Alyaa Mahmoud wrote:
>
> Hi All
>
> Is there a way to determine expression status of a gene
> (up, down) from
> corrected p.values ?
>
>
> No. The p-values are strictly between 0 and 1, so there is
no
> way to discern the sign of the underlying t-statistic. This
> would not be true if you were doing a one-sided t-test, but
I
> can't imagine a use case for a one-sided t-test in the
context
> of microarray analysis.
>
> Best,
>
> Jim
>
>
>
> Thanks a lot
> Alyaa
>
>
>
>
>
> --
> Alyaa Mahmoud
>
> "Love all, trust a few, do wrong to none"- Shakespeare
>
[[alternative HTML version deleted]]
Hi James
Thanks a lot, i thought so as well.
Its two sided
Thanks again
yours,
Alyaa
On Wed, Jul 11, 2012 at 4:21 PM, James W. MacDonald <jmacdon@uw.edu>
wrote:
> Hi Alyaa,
>
>
> On 7/11/2012 2:59 AM, Alyaa Mahmoud wrote:
>
>> Hi All
>>
>> Is there a way to determine expression status of a gene (up, down)
from
>> corrected p.values ?
>>
>
> No. The p-values are strictly between 0 and 1, so there is no way to
> discern the sign of the underlying t-statistic. This would not be
true if
> you were doing a one-sided t-test, but I can't imagine a use case
for a
> one-sided t-test in the context of microarray analysis.
>
> Best,
>
> Jim
>
>
>
>
>> Thanks a lot
>> Alyaa
>>
>>
> --
> James W. MacDonald, M.S.
> Biostatistician
> University of Washington
> Environmental and Occupational Health Sciences
> 4225 Roosevelt Way NE, # 100
> Seattle WA 98105-6099
>
>
--
Alyaa Mahmoud
"Love all, trust a few, do wrong to none"- Shakespeare
[[alternative HTML version deleted]]
