Cufflinks performs 0 splicing and promoter tests (Arabidopsis)

[dalesan]

Hello SEQAnswers community!

Thank you for this wonderful community here and to those of who you donate your time to checking forum posts and helping those in need.. I'm hoping that someone might know a thing or two as to where I can look to troubleshoot a particular problem that I'm having.

I should say that there has been at least 3 other people with the same problem, albeit with different organisms.

My issue: I cannot get cuffdiff v2.1.1 to perform any splicing and promoter preference tests. I couldn't get the previous version of cuffdiff to do so either...

My pipeline is as follows: Map with TrueSight -> cufflinks -> cuffmerge -> cuffdiff

Here's the command line for cuffdiff:

cuffdiff -L OX,KO,WT -o cuffdiffed --FDR 0.1 -b bowtie/genome.fa -N -c 5 cuffmerged/merged.gtf truesight_OX_out/alignment.bam truesight_KO_out/alignment.bam truesight_WT_out/alignment.bam

I'm using the iGenomes gtf file and bowtie files for all analyses.

Here's the cuffdiff stdout:

> Processed 30760 loci.                        [*************************] 100%
Performed 85169 isoform-level transcription difference tests
Performed 66792 tss-level transcription difference tests
Performed 61458 gene-level transcription difference tests
Performed 63795 CDS-level transcription difference tests
Performed 0 splicing tests
Performed 0 promoter preference tests
Performing 0 relative CDS output tests
Writing isoform-level FPKM tracking
Writing TSS group-level FPKM tracking
Writing gene-level FPKM tracking
Writing CDS-level FPKM tracking
Writing isoform-level count tracking
Writing TSS group-level count tracking
Writing gene-level count tracking
Writing CDS-level count tracking
Writing isoform-level read group tracking
Writing TSS group-level read group tracking
Writing gene-level read group tracking
Writing CDS-level read group tracking
Writing read group info
Writing run info

Here's the first 10 lines of my splicing.diff file (promoters.diff looks about the same)

test_id	gene_id	gene	locus	sample_1	sample_2	status	value_1	value_2	sqrt(JS)	test_stat	p_value	q_value	significant
TSS1	XLOC_000001	ANAC001	1:3630-5899	OX	KO	NOTEST	0	0	0	0	1	1	no
TSS10	XLOC_000007	GIF2	1:72338-74737	OX	KO	NOTEST	0	0	0.0384659	0	0.5459	1	no
TSS100	XLOC_000083	PGP11	1:523650-528745	OX	KO	NOTEST	0	0	0	0	1	1	no
TSS1000	XLOC_000884	LSH8	1:5785002-5785927	OX	KO	NOTEST	0	0	0	0	1	1	no
TSS10000	XLOC_009107	AT2G13910	2:5837685-5839920	OX	KO	NOTEST	0	0	0	0	1	1	no
TSS10001	XLOC_009108	AT2G13920	2:5844028-5845726	OX	KO	NOTEST	0	0	0	0	1	1	no
TSS10002	XLOC_009109	AT2G13940	2:5852758-5857264	OX	KO	NOTEST	0	0	0	0	1	1	no
TSS10003	XLOC_009110	AT2G13950	2:5857535-5859249	OX	KO	NOTEST	0	0	0	0	1	1	no
TSS10004	XLOC_009111	AT2G13960	2:5859584-5861311	OX	KO	NOTEST	0	0	0	0	1	1	no

In a previous post, a user was able to get the splicing tests to work by rolling back to an older version of cufflinks (version 1.30) -- but given that there have been so many improvements to the software, I don't want to do this..

So at this point, I humbly ask if anyone might know how to remedy this? Is it a bug? Is there something about Arabidopsis that doesn't allow splicing tests to be completed?

Thank you immensely for your time and advice, I assure you, it is sincerely appreciated!!

[nicrob]

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."

It looks like you only have one replicate for each condition, so you would need to change the --min-reps-for-js-test option to get any test output for splicing and promoter tests (keeping in mind the problem of too many false positives).

[dalesan]

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."

It looks like you only have one replicate for each condition, so you would need to change the --min-reps-for-js-test option to get any test output for splicing and promoter tests (keeping in mind the problem of too many false positives).

Nicrob, you're a genious. Thank you for bring this to my attention!!

[Hagen]

cuffdiff with 3 replica: but no splicing.diff output

Dear all,

- I am using cuffdiff v2.1.1 with 3 replicas of 3 samples

- the command line is:
cuffdiff --num-threads 4 --FDR 0.01 -o tmp --labels X,Y,Z annotation.gtf X_1.bam,X_2.bam,X_3.bam Y_1.bam,Y_2.bam,Y_3.bam Z_1.bam,Z_2.bam,Z_3.bam

- the resulting gene_exp.diff files contain for each gene 3 tests (one for each pairwise comparison)

- the files promoters.diff and splicing.diff however contain only a comment line and nothing more. From the previous posts, I was under the impression that with three replicas, these tests would be performed automatically. Am I wrong or missing anything else ?


Thank you