cufflinks: reports non existing transcripts

[kenietz]

Hi guys,
i just discovered a very strange thing.

Im using TopHat v2.0.5, cufflinks v2.0.2, samtools-0.1.18. Ref genome and GTF from Ensembl.

I have some PE RNA-seq so i mapped it with tophat and then used cufflinks. But cufflinks gave me some strange results. I got some transcripts which are nowhere to be found in the accepted_hit.bam file. And some of them have FPKM which is huuuuuge. But most have ok FPKMs.
------------------------------------
Mapped with:

tophat -o tophat_out -p 4 --b2-very-sensitive -r 150 -i 30 -I 700000 --coverage-search -G Tetraodon_nigroviridis.gtf Tetraodon_nigroviridis reads1.fastq reads2.fastq

Cufflinks command:

cufflinks -o cuffl_out -p 8 -g Tetraodon_nigroviridis.gtf -b Tetraodon_nigroviridis.fa -u -N -I 700000 tophat_out/accepted_hits.bam

Example from the transcripts.gtf:

Un_random Cufflinks transcript 45206342 45206435 1000 - . gene_id "cuff.305"; transcript_id "ENSTNIT00000024098"; FPKM "3567004609890.1142578125"; frac "1.000000"; conf_lo "3393898897531.518066"; conf_hi "3740110322248.710449"; cov "544435.333499"; full_read_support "yes";
Un_random Cufflinks exon 45206342 45206435 1000 - . gene_id "cuff.305"; transcript_id "ENSTNIT00000024098"; exon_number "1"; FPKM "3567004609890.1142578125"; frac "1.000000"; conf_lo "3393898897531.518066"; conf_hi "3740110322248.710449"; cov "544435.333499";


samtools view -F 4 tophat_out/accepted_hits.bam Un_random:45206342-45206435 , returns nothing. So no mapped reads to that region.
----------------------------------------------

So i have 3 questions:

q1: Why/how is this happening?
q2: How to avoid that situation?
q3: What should i do with these type of fake transcripts? Just filter them out?

Thank you for your time and help!

[iorikun]

Hi,

I have the same problem and I don't know what to do...