p*******r
发帖数: 4048 | 1
simple:
P
10-4.
You should use P-value. However, there is a thing called multiple hypothesis
correction. Basically, because you are testing for many genes, you can commit
an error just because of random fluctuations. As rule of thumb, divide 0.05 by
the number of genes on your chip.
in
the
Look them up in ensemble or something. It depends on what particular
biological question in mind.
Typically you want to confirm the top few ones with PCR. | l*****k
发帖数: 587 | 2 1. what algorithm did you use to calculate the P value? Each method may
give a different P value magnitude.
2. Affymetrix netaffyx has a GO tool, it is kinda useful.
3. It is very hard to publish in good journals with ONLY microarray
data now. You need to do more work to confirm your array data. I
personally don't belive in PCR confirmation.
4. I would suggest you do a permutation based analysis, one tool you can
use is ArrayTools from NCI Biometrix group.
Good Luck | f**r
发帖数: 70 | 3 First of all, if you could find a biostatistician in your department or your
school you can let him/her do the analysis for you. All that you are asking
is standard microarray data analysis.
You should use the p-values instead of fold changes in defining differentially
expressed genes. What puppeteer mentioned (0.05/N) is called Bonferroni
adjustment for multiple comparison, which is the most conservative way
possible. Other common and less-conservative way is FDR (false-discovery
rate) adjus | l*****k
发帖数: 587 | 4 I may be too 武断 in my previous post, however, here are my arguments.
with enough replicates, which is essential for any good microarray design. The
result is in most cases trustworthy. I think doing real time PCR to confirm is
useful in only several cases
1. gene of interest, and not so clearly shown change in array results.
2. want to know the exact fold change or abundance change.
there could be other reasons for doing this... like I said, this
"confirmation" will have to be put in a bigger sc
【在 p*******r 的大作中提到】
:
: simple:
: P
: 10-4.
: You should use P-value. However, there is a thing called multiple hypothesis
: correction. Basically, because you are testing for many genes, you can commit
: an error just because of random fluctuations. As rule of thumb, divide 0.05 by
: the number of genes on your chip.
: in
: the
| l*****k
发帖数: 587 | 5 permutation based test DOES NOT assume normal distribution of data as a lot of
statistics algorithms do.
and it incorporates probability theory into data analysis.
It works by randomly labeling your samples with the categories you do analysis
on, then do a T test (or any test). After doing so many of these things, it
will compared the T test results with the "supposed to be correct" labeling T
test result. and gives you a permutation P value.
by doing so, it actually is the most powerful in elim | f**r
发帖数: 70 | 6 It's indeed hard to argue. Irizarry wrote a paper to compare the 3 algorithms
in 2003 on Nucleic Acids Res. We've done some simulation studies and found
that his conclusions are pretty objective. While RMA is better with the
lower-expressed genes, it underestimates fold changes in a great deal, which
often times is very annoying to explain to the clinicians. But it's
interesting that in Affymetrix's to-be-released new algorithm (not called
MAS6, but some new name), they inherit the quantile
【在 l*****k 的大作中提到】
: permutation based test DOES NOT assume normal distribution of data as a lot of
: statistics algorithms do.
: and it incorporates probability theory into data analysis.
: It works by randomly labeling your samples with the categories you do analysis
: on, then do a T test (or any test). After doing so many of these things, it
: will compared the T test results with the "supposed to be correct" labeling T
: test result. and gives you a permutation P value.
: by doing so, it actually is the most powerful in elim
|
|
