Cufflinks FPKM inconsistent with reads in IGV

[lewewoo]

I have 6 samples the FPKM is as following:
0 0 0.23 0.15 0.56 0.3

however, when I used IGV to double checked the reads mapped, there are a lot of reads and the amounts mapped in the loci are pretty same; the total reads are not very different.

Any suggestions?

[Dario1984]

All those FPKMs are very low. Do you have any better examples ? I've seen this before in my data.

The FPKM estimates make no sense. For one, the second transcript from the top has no reads spanning the last intron. I scrolled all the way to the bottom. But it has a FPKM of 53.

I'm also not sure why the bottom-most transcript has a FPKM of 272. There are very few reads in the exonic part that is unique to it.

In summary, you don't want to trust Cufflinks estimates. They make no sense. Have you tried NEUMA instead ?

[hugq]

I found NEUMA work great and the results do make sense at many genes mannually checked.

All those FPKMs are very low. Do you have any better examples ? I've seen this before in my data.

The FPKM estimates make no sense. For one, the second transcript from the top has no reads spanning the last intron. I scrolled all the way to the bottom. But it has a FPKM of 53.

I'm also not sure why the bottom-most transcript has a FPKM of 272. There are very few reads in the exonic part that is unique to it.

In summary, you don't want to trust Cufflinks estimates. They make no sense. Have you tried NEUMA instead ?