I saw this thread, though it is older I just want to make sure I did my analysis after the update.

http://www.mail-archive.com/galaxy-user@lists.bx.psu.edu/msg00084.html

You mention having to repeatedly merge cufflinks output for cuffcompare to generate the combined transcripts. I was able to add them all in the cuffcompare section with the "add new Additional GTF input file". also my command is:

cuffcompare -o cc_output -r /galaxy/main_database/files/002/048/dataset_2048606.dat -R -s /galaxy/data/hg19/sam_index/hg19.fa ./input1 ./input2 ./input3 ./input4 ./input5 ./input6 ./input7 ./input8 ./input9 ./input10

This is correct: Galaxy now supports an arbitrary number of input files to cuffcompare, making repeated merging unnecessary.

Also I was reading on seqanswers that cuffdiff didn't work with biological replicates

http://seqanswers.com/forums/archive/index.php/t-7905.html, but when I did my galaxy analysis it had the "perform replicate analysis" available and was wondering if you had an updated version or did something else. I am analyzing a control and diseased group, with 5 biological samples in each group. Cuffdiff should work fine on this right? I got FPKM as output. Also do you know what statistical methods cuffdiff is using for analysis?

Cuffdiff claims to support biological replicates, and, as you found, Galaxy provides the infrastructure to specify replicates to Cuffdiff. I do not know the statistical methods used for analysis, but this would be a good question to ask either on seqanswers or on the new tophat-cufflinks mailing list: tophat.cufflinks@gmail.com

Lastly, in cuffdiff, the CDS, promoters and splicing analyses didn't seem to work (datasets 190,191,192). any ideas?

This is because the p_id attribute is not being attached to the combined_transcripts dataset produced by Cuffcompare. This problem is due to your reference file being incorrect -- it should be dataset 81, your reference annotation.

Thanks,