Negative RPKM values by EDGE-pro

[hi-koike]

Hello,

I am using your EDGE-pro programs for RNA-seq analysis
of bacterial species.
I could run the program, but I got negative values for some of
the samples.

I found rRNA.numberReads was more than numberReads or
numberUniqueReads, thus I got negative value for
rpkm.numberReads.

I tried to find the reason by reading the perl script, but I found
counting of the number of reads was done by a binary code.

I would appreciate it if you could tell me why and how to
do to run the program.

Thanks in advance,
Hideaki

[dpryan]

Why don't you send an email to the address listed on the EDGE-pro webpage?

[hi-koike]

Hi dpryan,

Thank you for your advice. Actually I did, but it didn't work.
I would appreciate any idea.

Thank you,
Hideaki

[blakeoft]

The only thing I can think of is that the RPKM values are log transformed, but the manual doesn't seem to mention that they're log transformed. When did you try emailing them? It's possible that they haven't had a chance to reply yet.

[hi-koike]

Hi blakeoft,

Thank you. As you pointed, I will wait the answer for my email.

I tried to read the perl script, but the count was done by
a binary code "count". So I need a help from a well experienced person.

I have waited only for 3 days, so I will wait.

Thank you,
Hideaki

[dpryan]

You might have to give it a few weeks. Many authors are pretty bad at getting back to people in a timely manner.

[freestile]

Hi blakeoft,

Thank you. As you pointed, I will wait the answer for my email.

I tried to read the perl script, but the count was done by
a binary code "count". So I need a help from a well experienced person.

I have waited only for 3 days, so I will wait.

Thank you,
Hideaki

Finally you solved ?? I have the same problem.
Regards

[dbouzo]

I am also experiencing this same problem, however my .numberReads file (total number of reads) is a value smaller than that of the rRNA.numberReads file (number of reads mapped to rRNAs). I don't understand how this is possible as I would interpret 'total number of mapped reads' as including the reads from rRNA.

Has anyone else had this issue at all, or an update with a way to resolve this? I have contacted the EDGE-pro developers and hope to hear back soon.

[Markiyan]

Try removing rRNA, tRNA, transposon reads.

In order to minimise the impact of the bug, try removing the reads matching rRNA, tRNA and transposase genes present in a genome of the analysed organism and rerunning the mapping/analysis.

Also check the definition of the rRNA regions (how the program gets those reads numbers).

So first make sure you have a multifasta file with rRNA (full rrn loci) and tRNA seqs for your organism, than you can use my tool as a simple contaminant remover:
(assuming adapters had been clipped already):

fastq_miseq_trimmer [R1.fastq.gz] [R2.fastq.gz] -base=[output_files_basename] -vs=[rtRNAs.fasta] -nohproc -noclip

Other processing options:
To remove illumina adapters: replace -noclip with -ts for Illumina TruSeq adapters clipping. To set quality clipping threshold add: -epc=0.5 (0.5=cumulative error probability in 10bp sliding window).

To set the ends to /1 /2, If your tool wants /1 or /2 as ends specifier, and replace ":" in a read_name with "_"
replace -nohproc with -ends=3

http://seqanswers.com/forums/attachment.php?attachmentid=4343&d=1463489447

[dmking]

hi-koike was there a resolution of this issue? Or did you use the solution suggested by Markiyan?

I'm having the same problem and have emailed the authors but would like to know what, if any, approach worked for you.

[tdcollingsworth]

Checking In

Hey, guys -

Pretty stumped at this as well: am in the process of trying to debug 'edge.pl' as well as 'count.cpp'.

The code is good but is pretty convoluted and not well documented: pretty standard scientific code, I suppose.

Interested in hearing about anyone else's success in figuring out where the bug is!

Cheers,

Timothy