On 8/21/12 4:33 AM, i b wrote:
Thanks Jen, useful link. But I did not understand one thing. I have the following FPKM in cufflinks for two samples: s1 (untreated): 1234106 s2 (treated): 159713
cuffdiff of the two samples gives me the following values: value_1: 5.4 value_2: 20.9
and it is not significant (!). My two question:
- how is this not significant?
- what is the realtion between the high fpkm in cufflinks and the low
values in cuffdiff?I read the manual: is this part of the statistical method adopted?e.g are these numbers (cuffdiff values) derived from the formula adopted?
thanks a lot, ib
On Thu, Aug 16, 2012 at 11:26 PM, Jennifer Jackson jen@bx.psu.edu wrote:
Hello,
A very similar question came up a few days ago and Jeremy had some good advice for how to approach learning to interpret this data:
http://lists.bx.psu.edu/pipermail/galaxy-user/2012-August/004985.html
Best,
Jen Galaxy team
On 8/15/12 8:49 AM, i b wrote:
Dear all, in cuffdiff outputs e.g. transcript differential expression, I find for example: value_1 value_2 log2(fold_change) 7.77183 0 -1.79769e+308
or
value_1 value_2 log2(fold_change) 0 14.5972 1.79769e+308
for many many rows.
if I sort in excel my data by fold change column (big to small ), all the rows with -1.79769e+308 or +1.79769e+308 are on the top. How can be sure that these on the top are really the most up-regulated or down regulated transcripts if I don't know the real value of one of the two samples (is 0 really zero?)? I was told that the zero in one if the two samples is very small number and Cuffdiff simply writes 0, but it is not absolutely zero, otherwise it would not be possible ot have -1.79769e+308 or 1.79769e+308
Could you please tell me then how can I extrapolate the highest fold change? (up and down regualted)?or of what is done by sorting by log fold chnage is correct?
Thanks, ib ___________________________________________________________ The Galaxy User list should be used for the discussion of Galaxy analysis and other features on the public server at usegalaxy.org. Please keep all replies on the list by using "reply all" in your mail client. For discussion of local Galaxy instances and the Galaxy source code, please use the Galaxy Development list:
http://lists.bx.psu.edu/listinfo/galaxy-devTo manage your subscriptions to this and other Galaxy lists, please use the interface at:
http://lists.bx.psu.edu/-- Jennifer Jackson http://galaxyproject.org