NOTEST everywhere with cuffdiff

[ramiro2k]

Hello, I am running the following pipeline:

tophat -p 4 -G genes.gtf -o sample1TopHat genome sample1R1.fastq sample1R2.fastq
tophat -p 4 -G genes.gtf -o sample2TopHat genome sample2R1.fastq sample2R2.fastq

cufflinks -o sample1Cufflinks -p 4 -G genes.gtf sample1TopHat/accepted_hits.bam
cufflinks -o sample1Cufflinks -p 4 -G genes.gtf sample1TopHat/accepted_hits.bam

cuffmerge -o sampleCuffMerge -p 8 -g genes.gtf -s genome.fa assemblylist.txt

cuffdiff -o sampleCuffdiff -p 8 -b genome.fa -L s1,s2 -u sampleCuffMerge/merged.gtf sample1TopHat/accepted_hits.bam sample2TopHat/accepted_hits.bam

However, I am getting NOTEST on the splicing.diff, whereas the isoform_exp.diff does show some information. I am just not sure how to debug this. Any suggestions?

I have messed with the -c flag in cuffdiff. I am using cufflinks 2.2.1, and I am using the UCSC Homo Sapiens HG19 genome and annotation (genes.gtf). Any suggestions appreciated.

Thank you in advance,
Ramiro

[Bukowski]

So, you're expecting significant results from an experiment with no replicates?

Also see : https://www.biostars.org/p/163535/

[ramiro2k]

Hello,

Thank you so much for your reply. I ran it with two replicates and got a similar result, nothing significant, nothing expressed in the splicing.diff. I had looked at the link you mentioned in the past but I am surprised that nothing comes out as expressed or significant. On the differentialy expressed results I do get many genes as significant, etc. I am guessing this means that no splicing events ocurred, yet if you or anyone has any suggestions about anything I might be doing wrong please let me know.

Thank you,
Ramiro

tophat -p 4 -G genes.gtf -o sample1r1TopHat genome sample1r1R1.fastq sample1r1R2.fastq
tophat -p 4 -G genes.gtf -o sample1r2TopHat genome sample1r2R1.fastq sample1r2R2.fastq
tophat -p 4 -G genes.gtf -o sample2r1TopHat genome sample2r1R1.fastq sample2r1R2.fastq
tophat -p 4 -G genes.gtf -o sample2r2TopHat genome sample2r2R1.fastq sample2r2R2.fastq

cufflinks -o sample1r1Cufflinks -p 4 -G genes.gtf sample1r1TopHat/accepted_hits.bam
cufflinks -o sample1r2Cufflinks -p 4 -G genes.gtf sample1r2TopHat/accepted_hits.bam
cufflinks -o sample2r1Cufflinks -p 4 -G genes.gtf sample2r1TopHat/accepted_hits.bam
cufflinks -o sample2r2Cufflinks -p 4 -G genes.gtf sample2r2TopHat/accepted_hits.bam

cuffmerge -o sampleCuffMerge -p 8 -g genes.gtf -s genome.fa assemblylist.txt

cuffdiff -o sampleCuffdiff -p 8 -b genome.fa -L s1,s2 -u sampleCuffMerge/merged.gtf sample1r1TopHat/accepted_hits.bam,sample1r2TopHat/accepted_hits.bam sample2r1TopHat/accepted_hits.bam,sample2r2TopHat/accepted_hits.bam

[ramiro2k]

I actually found the answer in another post from 2011:

One of the updates going from version 1.3 to version 2 (quoting from the cufflinks website here) was:

"After substantial performance testing, we have determined that the false discovery rate of Jensen-Shannon-based tests (differential splicing, CDS switching, and promoter switching) can be unnacceptably high when used with fewer than three replicates in the conditions being compared. Cuffdiff now refrains from performing significance tests when one of the conditions involved has fewer than three replicates. You can change this behavior with the new --min-reps-for-js-test option. Cuffdiff still produces splicing.diff, cds.diff, and promoters.diff regardless of how many replicates you have. These files will include the JS distance scores, but none of the genes will be marked significant if you have fewer than the required number of replicates."

This explained the situation.