FPKM values are zero

[budgie lover]

Hi everyone,

When I run Cuffcompare on the merged.gtf I get from Cuffmerge, the FPKM values are zero in the tmap. Does anyone have any idea why this might happen? Running Cuffcompare on individual the individual tissues gives me FPKM values. I did not get an errors in when running anything except when I ran Cuffmerge I got:

[bam_header_read] EOF marker is absent. The input is probably truncated.
[bam_header_read] invalid BAM binary header (this is not a BAM file).
File set_1_merge/tmp/mergeSam_fileHkhL9l doesn't appear to be a valid BAM file, trying SAM...

But Cuffmerge completed without any errors. I'm not really sure what file Cuffmerge is looking for. Any information on how Cuffmerge works is greatly appreciated.

Thank you.

[Chirag]

I am not exactly sure, but my best guess would be, When you give a reference GTF, while doing cufflinks/cuffmerge, if there are some RNASeq raw reads around those GTF trasncript, it (SEEMS TO) output those transcript.

To check this: Take those trasncript from "genes.fpkm_tracking" where all FPKM values are 0. Now go these regions and have a look at the raw reads on the browser. You will see (atleast i do in my data) there are some raw reads mapping around those trasncripts, but cufflink is not able to reconstruct the transcript, let alone on it own

So for now, i extract all those region and exclude them out from the merged.gtf, so i can get the transcript that is actually build on data that we are working.

This is how i fell, i am not sure, if i am correct !

cheers
CN