[adumitri]

Hi Thomas,

Can you be more specific?

From the Cuffdiff manual I understand that technical replicates are supported, but not biological replicates. For example, if I have 3 control and 4 disease RNA-Seq samples (all samples different), can I use Cuffdiff to analyze the differential expression between the control and the disease groups? If the answer is yes, how would I do it?

Cuffdiff manual:
Cufflinks includes a program, "Cuffdiff", that you can use to find significant changes in transcript expression, splicing, and promoter use. From the command line, run cuffdiff as follows:
cuffdiff [options]* <transcripts.gtf> <sample1_replicate1.sam[,...,sample1_replicateM]> <sample2_replicate1.sam[,...,sample2_replicateM.sam]>... [sampleN.sam_replicate1.sam[,...,sample2_replicateM.sam]]

[marcora]

Hi Thomas,

Can you be more specific?

From the Cuffdiff manual I understand that technical replicates are supported, but not biological replicates. For example, if I have 3 control and 4 disease RNA-Seq samples (all samples different), can I use Cuffdiff to analyze the differential expression between the control and the disease groups? If the answer is yes, how would I do it?

I would also like to know what difference it makes to cufflinks if the replicates are technical or biological, if any... and if any, why?

Cheers

[RockChalkJayhawk]

The replicates are the ones separated by commas (these should be biological replicates) and the control versus treated should be separated by spaces.

Although I don't understand the inner workings of CuffDiff, I do know that you need replicates of each group to assess the variability of your measurement. If you have a lot of variation within in your replicates, then the effect between groups will need to be much larger to call differential expression.

[Thomas Doktor]

Technically it doesn't matter if the replicates are biological or technical as far as cufflinks is concerned (from my understanding at least), but the variation that you estimate based on the replicates will of course be affected. With biological replicates you will have contribution to the variation from both the inherent biological variability as well as the technical variability.
In terms of calling genes as differentially expressed, biological replicates are the safer way to go about it since you are less likely to get false positives. What actually constitutes a biological replicate is another matter (same cell-line but different preparation, different cell-line, etc).

[adumitri]

Thank you both for your help!

It would be great if anyone from the Cuffdiff team would be able to comment on how exactly the program decides on differential expression between groups. I am not sure I understand the process from what is available on the Cufflinks website. Is there any additional documentation?

Alexandra

[RockChalkJayhawk]

Thank you both for your help!

It would be great if anyone from the Cuffdiff team would be able to comment on how exactly the program decides on differential expression between groups. I am not sure I understand the process from what is available on the Cufflinks website. Is there any additional documentation?

Alexandra

I believe they will refer you to the supplemental information from thier paper, but I think they are coming out with a new paper because of all the new options and methods.

[Xi Wang]

I believe they will refer you to the supplemental information from thier paper, but I think they are coming out with a new paper because of all the new options and methods.

In their supplementary materials, it is not mentioned how to deal with groups of samples, or how to estimate the (technical and biological) replicate groups' variance on genes' expression levels. I agree the guess that a new publication will come out soon by their group, but I really appreciated that if someone (in the group) can give a general idea on how it works. Thanks.

[honey]

RNA seq analysis

Hi
I am using Galaxy for RNA seq analysis. Should I run the analysis in web based version (takes a long time) or download it locally. Any suggestion

[Xi Wang]

Hi
I am using Galaxy for RNA seq analysis. Should I run the analysis in web based version (takes a long time) or download it locally. Any suggestion

Cuffdiff really takes a long time when the sample size is large, no matter you run it online or locally.