Hi Maria, I didn't notice any obvious problematic usage, format, or content issues with the Tuxedo pipeline execution in your history. Your protocol is right on track. This leaves data and parameter inputs to consider. I did notice that you are mainly using defaults and omitting the use of reference annotation that Cuffdiff uses to generate the full compliment of statistics. The "NOTEST" result indicates that the coverage is too shallow. You could follow the advice here, by adjusting "-c" to be lower. This is "Min Alignment Count:" and is set to "10" in your runs. http://cufflinks.cbcb.umd.edu/faq.html#notest Adding in a reference annotation file could also potentially help. Aligned sequences may be falsely fragmenting without a reference transcript to help bind them together. But, this is just a guess - I didn't examine any assembly regions. This is however something that you could do. The UCSC Table Browser is one source for a GTF file. Experimenting with other parameters as you are doing also is worth it. The manual and such cover these in detail, and there is always the tool author's google group for detailed questions/advice. Good luck with your project, Jen Galaxy team On 2/6/14 12:38 PM, Maria Hoffman wrote:
Hello,
Thank you for your help. I have found that wiki page very helpful and actually us it very often (I was using it this AM too before I emailed you). In looking at the wiki again, nothing is really standing out to me ( my chromosome notation matches up etc). I am going to keep looking etc but I did send you my history too. I did try running another cuffdiff playing with the dispersion estimation method too out of curiosity.
Thank you so much for your help! This is my first real data set doing this and we have abstracts due soon, so the pressure is on!
Thanks! Maria
-- Jennifer Hillman-Jackson http://galaxyproject.org