very high RPKM values from Cufflink

[yumtaoist]

Is your reference very short?

[pbluescript]

honey, it might be a good idea to look a bit more in depth into that specific gene. You can certainly get high FPKMs mapping to genes like actin that make up a lot of the mRNA percentage of a cell. I had huge numbers of reads mapping to one region of a miRNA gene once that all turned out to be within a LINE and a SINE. For that gene at least, it was clear the repeat regions skewed the results.

[honey]

High RPKM

Thanks I looked at the Bam files and can say that there are very few reads wherver it is 0 values of RPKM however where the values are very high those are the kind of hot spots there are large no of reads. Now the question is is this an artifact -High RPKM or very low RPKM how we rope up both extreme values?

[peromhc]

Honey, I could be totally wrong here about the log(10) thing, but I don't think I am..

Can you look at the mappings for some of those transcripts where 'raw' FPMK is about 0-- do they have few reads mapped?

See:

http://seqanswers.com/forums/showthread.php?t=16962

[honey]

very high RPKM values from 4.5 to sevreal thousands

Howver the problem which I have is that the RPKM are high apx. 5% are > 1000RPKM (like 3245, 4356 and so on) in the same sample. If I change to log 10 than then what will happen to values around zero. Is it a usual method to log transform RPKM value?
Any feedback is welcome
Thanks

[peromhc]

I ran an RNA-seq experiment and used TopHat> cufflink It is a time series experiment when I looked at RPKM values in some of the transcripts the RPKM values goes upto 32765.04908; 2073.978485 . Is is reasonable? I also looked at Bam files there weresecrtainly very large number of reads. Any feedback/ suggestion please how to explain this high RPKM values?

I think that these values should be taken to the log(10).. this is not documented, but my suspicion.

log(10) values from cufflinks roughly equals FPKM values from cuffdiff..

[honey]

I ran an RNA-seq experiment and used TopHat> cufflink It is a time series experiment when I looked at RPKM values in some of the transcripts the RPKM values goes upto 32765.04908; 2073.978485 . Is is reasonable? I also looked at Bam files there weresecrtainly very large number of reads. Any feedback/ suggestion please how to explain this high RPKM values?