Cufflinks 0 FPKM values

[herstein]

Hi,

We are running tophat/cufflinks on 3 identical human rna samples which were sequenced in 3 different lanes on 2 different flowcells. All 3 samples are from the exact same library.

When running cufflinks, the genes.fpkm_tracking file for one sample is showing some genes with 0 FPKM and status FAIL, however these same genes in the other two samples are showing a non-zero FPKM.

We are using tophat-1.4.0, bowtie-0.12.7, and cufflinks-1.3.0

We ran tophat with the -G option and provided the gtf file. We ran cufflinks with -G option providing the same gtf file that we gave to tophat. We also ran cufflinks with -u -N --compatible-hits-norm -b hg19_ucsc.fa

We've tried running without options -N and --compatible-hits-norm but still had the same results.

We are confused how with 3 identical samples, one shows up with 0 FPKM but the other two seem to have valid FPKM's. For the 0-FPKM gene, we checked the bam file in IGV and it is definitely mapped.

Has anyone encountered a similar problem or does anyone have any insight into this?

Thanks!
--Jennifer

[letusgo]

Hi,
Have you solved this problem yet? I met the same problem now.

[jp.]

I also have same problem. I got only few genes with 0 FPKM value.
I mean, control vs treated; control has 0 FPKM but treated shows 1098.13 FPKM.
Theoretically speaking, it has very high fold increase, but cuffdiff Inf ?????
do I have to include it or exclude. I think, this is very significant expression ????
Can you all tell me please ?