with cufflinks, can there be "0" fpkm value and yet full read support?

velu · 04-21-2017, 10:05 PM

Hi,
I used hisat2 (2.0.5) to map 150b Paired-end reads on to the draft genome of castorbean and used cufflinks (version 2.2.1) to perform reference annotation based transcript assembly (RABT). This data is NOT for differential expression analysis. The assembler reported novel isoforms and some novel genes. From the results, of the transcripts that were as present in the gtf file used for RABT, some had an FPKM value of "0.0000000000" but were "yes" for "full_read_support" whereas the rest that were "yes" for "full_read_support" carried an FPKM value at 1 or above.
Could anyone please explain why would a transcript be described as present with "full_read_support," yet with an FPKM value of "0.0000000000?"

Thanks.

Dario1984 · 04-24-2017, 08:00 PM

I have noticed similar problems with Cufflinks in the past. Our research group no longer uses it because it hasn't been updated in 3 years and some of the FPKM values look suspicious, which you have noticed.

velu · 04-25-2017, 05:31 AM

Thank you, Dario1984.
Then, what method do you follow for the assembly of reads that do not have such limitations?

Dario1984 · 04-25-2017, 07:00 PM

We use Trinity because we found its results can be experimentally validated by biologists. There's also a genome guided assembly mode. It was developed a few years ago, but it still maintained by the developer. You might also be interested in categorising your assembled transcripts with TransDecoder.

velu · 04-25-2017, 07:23 PM

Thank you.
I will go through the possibilities you have suggested.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
DESeq2: bad experimental design and "not full rank" problems	Erinyes	Bioinformatics	2	09-15-2015 02:40 AM
DESeq2 paired data: "model is not full rank"	sbcn	Bioinformatics	5	09-14-2015 12:03 PM
Bcl2Fastq 1.8.3 install error on Ubuntu 12.04LTS: "No support for bzip2 compression"	slees	Bioinformatics	3	10-26-2013 02:52 AM
How does cuffdiff/cufflinks determine "read type"?	sdarko	Bioinformatics	3	09-01-2011 12:13 AM
calibrated quality string "B" full	Trudy	Bioinformatics	2	03-25-2011 11:04 AM

04-21-2017, 10:05 PM	#1
velu Junior Member Location: Hyderabad, India Join Date: Sep 2015 Posts: 4	with cufflinks, can there be "0" fpkm value and yet full read support? Hi, I used hisat2 (2.0.5) to map 150b Paired-end reads on to the draft genome of castorbean and used cufflinks (version 2.2.1) to perform reference annotation based transcript assembly (RABT). This data is NOT for differential expression analysis. The assembler reported novel isoforms and some novel genes. From the results, of the transcripts that were as present in the gtf file used for RABT, some had an FPKM value of "0.0000000000" but were "yes" for "full_read_support" whereas the rest that were "yes" for "full_read_support" carried an FPKM value at 1 or above. Could anyone please explain why would a transcript be described as present with "full_read_support," yet with an FPKM value of "0.0000000000?" Thanks.

04-24-2017, 08:00 PM	#2
Dario1984 Senior Member Location: Sydney, Australia Join Date: Jun 2011 Posts: 166	I have noticed similar problems with Cufflinks in the past. Our research group no longer uses it because it hasn't been updated in 3 years and some of the FPKM values look suspicious, which you have noticed.

04-25-2017, 05:31 AM	#3
velu Junior Member Location: Hyderabad, India Join Date: Sep 2015 Posts: 4	Thank you, Dario1984. Then, what method do you follow for the assembly of reads that do not have such limitations?

04-25-2017, 07:00 PM	#4
Dario1984 Senior Member Location: Sydney, Australia Join Date: Jun 2011 Posts: 166	We use Trinity because we found its results can be experimentally validated by biologists. There's also a genome guided assembly mode. It was developed a few years ago, but it still maintained by the developer. You might also be interested in categorising your assembled transcripts with TransDecoder.

04-25-2017, 07:23 PM	#5
velu Junior Member Location: Hyderabad, India Join Date: Sep 2015 Posts: 4	Thank you. I will go through the possibilities you have suggested.