why is FPKM equal to 0 in transcripts.gtf generating by cufflinks?

Question

[512788522]

I have analyzed some RNA-seq data with Tophat and cufflinks, and I have several problems about the output. I run the tophat and cufflinks with default values, that is, I used the following command:

$/opt2/tools/tophat-2.0.13.Linux_x86_64/tophat -p 5 -G genes.gtf -o tophat_mut ./ucsc.hg19 2-K13-mut
.fastq.gz
$/opt/toolkit/cufflinks-2.2.1.Linux_x86_64/cufflinks -p 5 -u -g genes.gtf -o ./cufflinks ./tophat/a
ccepted_hits.bam

But the output (i.e. transcripts.gtf) looked like strange.

chr1    Cufflinks    transcript    34611    36081    1    -    .    gene_id "FAM138A"; transcript_id "NR_026818_1"; FPKM "0.0000000000"; frac "0.000000"; conf_lo "0.000000"; conf_hi "0.000000"; cov "0.000000"; full_read_support "no";
chr1    Cufflinks    exon    34611    35174    1    -    .    gene_id "FAM138A"; transcript_id "NR_026818_1"; exon_number "1"; FPKM "0.0000000000"; frac "0.000000"; conf_lo "0.000000"; conf_hi "0.000000"; cov "0.000000";

chr1    Cufflinks    exon    140075    140566    1000    -    .    gene_id "CUFF.2"; transcript_id "NR_039983"; exon_number "3"; FPKM "0.0401424461"; frac "1.000000"; conf_lo "0.000000"; conf_hi "0.080304"; cov "0.037191";
chr1    Cufflinks    transcript    323892    328581    1    +    .    gene_id "CUFF.1"; transcript_id "NR_028322_1"; FPKM "0.0000000000"; frac "0.000000"; conf_lo "0.000000"; conf_hi "0.000000"; cov "0.000000"; full_read_support "no";

1.Why wad FPKM equal to 0? I found that many FPKM values were equal to 0, but some FPKM values were very high. Was this situation unusual? Why did it occur? 2.What did the gene_id "CUFF.2" (or gene_id "CUFF.1") mean?

Thanks!

Answer 1

[ashwini]

FPKM=0 means that there were no reads which mapped to that location. Which may also mean that the particular transcript was not expressed.

Answer 2

[Devon Ryan]

1. If no reads map to a gene, then its FPKM is 0.
2. It has no meaning. Cufflinks just sequentially numbers features.

[Dawn]

I have similar question.

Cufflinks has -F/--min-isoform-fraction option set to 10% by default so it should suppress isoforms of FPKM=0. But when I run Cufflinks with -g/--GTF-guide option, I usually get isoforms with FPKM=0 in output while there are other expressed transcripts in the same gene.

[Devon Ryan]

Cufflinks just doesn't try to tweak the assembly of those isoforms. The FPKMs should still be output.

Answer 3

[pengchy]

It seems the reason that you have used a reference gene annotation file. The reference gene will output with FPKM 0 when there is no reads hit on them. However, after having checked the reference gene number appeared in the transcripts.gtf, I found it is less than the supplied "-r" reference gene number. In my case, 17500 in the reference and 16900 in the transcripts.gtf file. Still confused to this phenomenon.

Anyone who can give a explanation will be very appreciated.

[pengchy]

http://seqanswers.com/forums/showthread.php?t=8218

This thread in seqanswers give more information and examples, but the problem still there.