I think the sign is to show if it is x-fold more than the first condition (+) or x-fold less than the first condition (-). A regular fold would give you values from 1-whatever if sample 2 is more than sample 1, and a fraction (0-1) if sample 1 is expressed more than sample 2. The log lets you get both on the same scale, so that 2 means (on log2 scale) four-fold upregulated, and -2 means four-fold downregulated.

Hi Jiwen,

As far as I know, this is possible. The CuffDiff log2 value is defined here: http://cufflinks.cbcb.umd.edu/manual.html#gene_exp_diff

7   FPKMx              8.01089   FPKM of the gene in sample x
8   FPKMy              8.551545     FPKM of the gene in sample y
9   log2(FPKMy/FPKMx)  0.06531     The (base 2) log of the fold
                                 change y/x

And log(2) in general (including a graph, which can help with visualizing) is described here (although the web is full if similar):
http://en.wikipedia.org/wiki/Logarithm

I can see that this same Q on seqanswers.com recieved pretty much the same answer (in brief! http://seqanswers.com/forums/showthread.php?t=19558), so I think you are safe using the data as it as generated. Taking a look at the inputs would be advised if the results were unexpected.

If you do still have concerns about the log(2) calculation, asking the tool authors directly (if you have't done so already) at tophat.cufflinks@gmail.com is always an option, to tripple check.

Best,

Jen
Galaxy team

On 4/26/12 7:48 AM, 杨继文 wrote:

Hi,
I am analyzing my RNA-Seq data. After running cuffdiff, I got a list of
differentially expressed transcrpts or genes.
As far as I know, log2 value = fold change. However, there are minus
values. Is this possible?? log2 value can not be minus. Did I miss
something??
Looking forward to your help.
Thanks in advance.
Best
Jiwen


------------------------------------------------------------------------
网易Lofter，专注兴趣，分享创作！ <http://www.lofter.com>


___________________________________________________________
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-dev

To manage your subscriptions to this and other Galaxy lists,
please use the interface at:

   http://lists.bx.psu.edu/