[Anelda]

Hi Cole,

I've created a subset of the data and thought I'd quickly run it to see how cuffmerge perform before sending it to you... And surely, cuffmerge seems to have worked fine - no errors. I'm re-running now with the full dataset and will let you know what happens.

[aeveland]

Hi Cole,
I have a question about the min-alignment-count option in CuffDiff. In previous versions I've played around with adjusting -c between 100 and 500 - the default was 500. Now in v1.0.1 the default is 10. Can you clarify - this is 10 paired-end reads mapping to a given locus across all samples being tested? Sounds quite low and a huge difference from previous versions. Am I missing something?
Also, what are your recommendations for this in general?
Thanks!
Andrea

[pinki999]

Hi Cole,

I have 2 sample groups each containing 5 biological replicates. I am using NCBI36 build annotation file and the sorted bam files. I am trying to use cuffdiff but am getting errors:

Error: cannot open alignment file for reading

And also the bam files are not opening.

What's going wrong here?

Thanks

[lshen]

In the new version, the FPKM is around 40000 time of that in v0.9.3. What's the reason of the inflated FPKM values in version 1.0.1?

## Version 1.0.1
tracking_id class_code nearest_ref_id gene_id gene_short_name tss_id locus length coverage status FPKM FPKM_conf_lo FPKM_conf_hi
C4.1 - - C4.1 - - chr1:28377-29113 - - OK 1.43906e+07 9.13587e+06 1.96452e+07
C4.2 - - C4.2 - - chr1:29315-30559 - - OK 9.59506e+06 6.54169e+06 1.26484e+07
C4.3 - - C4.3 - - chr1:30622-33290 - - OK 1.42618e+07 6.77115e+06 2.17525e+07
C4.4 - - C4.4 - - chr1:399114-408248 - - OK 8.96664e+06 6.40474e+06 1.15285e+07
C4.5 - - C4.5 - - chr1:382665-396076 - - OK 3.21014e+07 2.29313e+07 4.12715e+07
C4.6 - - C4.6 - - chr1:424707-428382 - - OK 3.70756e+07 1.65899e+07 5.75614e+07
C4.7 - - C4.7 - - chr1:415165-421285 - - OK 1.49276e+08 1.34216e+08 1.64336e+08
C4.8 - - C4.8 - - chr1:531526-535663 - - OK 2.10248e+07 1.39171e+07 2.81325e+07
C4.9 - - C4.9 - - chr1:522058-525313 - - OK 2.77026e+07 1.38614e+07 4.15439e+07

## Version 0.9.3
gene_id bundle_id chr left right FPKM FPKM_conf_lo FPKM_conf_hi status
C4.25 115831 chr1 28377 29113 346.747 309.504 383.989 OK
C4.41 115843 chr1 399114 408248 243.432 212.228 274.637 OK
C4.47 115842 chr1 382665 396076 851.888 793.513 910.262 OK
C4.55 115851 chr1 424718 428382 763.402 708.142 818.661 OK
C4.65 115832 chr1 29315 76807 229.826 199.173 260.479 OK
C4.67 115832 chr1 67645 67784 919.886 859.227 980.545 OK
C4.87 115847 chr1 415165 421449 3084.75 2973.67 3195.83 OK
C4.115 115879 chr1 531526 535663 484.381 440.363 528.398 OK
C4.123 115877 chr1 522058 525313 817.005 695.07 938.939 OK

[Cole Trapnell]

Are you using -N? If so, this is due to changes in the way we were doing the quantile normalization. In 1.0.1, we are dividing by the number of mapped reads in bundle (i.e. a gene) at the 75th percentile in the distribution of fragments per bundle. In 0.9.3, we were dividing by the sum of counts in all bundles at or below the 75th percentile. That's a much larger number. We are actually implementing another scaling factor (similar to what Bullard et al did in their paper) that should make the FPKM values you get with and without -N comparable. So basically, this isn't a bug- it's expected behavior. However, it's confusing enough to users (you're not the first to ask this) that we're going to make a few more tweaks in the upcoming 1.0.2 release to -N.

[kenosaki]

Hi, does anyone have experienced the error when running caffmerge?
I think merge file has no problem.
Cufflinks was run using a sorted bam file, and I have three cufflinks results.

And my command is this.
cuffmerge -g ../bowtie_work/bowtie_ref/refGene.gtf -s ../bowtie_work/bowtie_ref/hg19.fa assemblies.txt

regards,
Ken

This is the error message....

[Wed Jun 1 10:31:29 2011] Beginning transcriptome assembly merge
-------------------------------------------

[Wed Jun 1 10:31:29 2011] Preparing output location ./merged_asm/
[Wed Jun 1 10:31:36 2011] Converting GTF files to SAM
[10:31:36] Loading reference annotation.
[10:31:38] Loading reference annotation.
[10:31:40] Loading reference annotation.
[Wed Jun 1 10:31:42 2011] Quantitating transcripts
You are using Cufflinks v1.0.2, which is the most recent release.
[bam_header_read] EOF marker is absent.
File ./merged_asm/tmp/mergeSam_file5VS2mi doesn't appear to be a valid BAM file, trying SAM...
[10:31:42] Loading reference annotation.
[10:31:46] Inspecting reads and determining fragment length distribution.

Error: this SAM file doesn't appear to be correctly sorted!
current hit is at chr10:180985, last one was at chr1:249211330
Cufflinks requires that if your file has SQ records in
the SAM header that they appear in the same order as the chromosomes names
in the alignments.
If there are no SQ records in the header, or if the header is missing,
the alignments must be sorted lexicographically by chromsome
name and by position.

[FAILED]
Error: could not execute cufflinks

[kenosaki]

And my command is this.

cuffmerge -g ../bowtie_work/bowtie_ref/refGene.gtf -s ../bowtie_work/bowtie_ref/hg19.fa assemblies.txt

[Anelda]

Hi Kenosaki,

For related posts and possible solutions you can go to http://seqanswers.com/forums/showthr...ight=cuffmerge

[kenosaki]

Thanks Anelda, I'll check this thread...

[nsl]

Hi Cole,
I have a question about the min-alignment-count option in CuffDiff. In previous versions I've played around with adjusting -c between 100 and 500 - the default was 500. Now in v1.0.1 the default is 10. Can you clarify - this is 10 paired-end reads mapping to a given locus across all samples being tested? Sounds quite low and a huge difference from previous versions. Am I missing something?
Also, what are your recommendations for this in general?
Thanks!
Andrea

Hi Andrea (aeveland) and any one who has some insight,

I have the same question about option -c in cuffdiff. Why are the defaults so different and how do they translate. If you found the answer would you mind sharing it with me.
I have been teaching myself rna-seq analysis for a little under 2 months and am using Galaxy to get my feet wet. The same default option -c there for cuffdiff is set at a 1000.
what is a conservative setting any suggestions?

Isthuthiy = thanks

nsl

[lshen]

default option -c there for cuffdiff is set at a 1000

It is too high, now the default is lowered.

Galaxy maybe still using older version.

[ikaka89]

Hi,

How can I get the number of unmapped reads from the cufflinks output. I have a bam file which was generated from tophat. I converted it into a sam file, There were 38 million lines in the samfile and all of them are mapped according to the flag(in the samfile), but when I run cufflinks the total mapped mass is only 32 million. does that mean that 6 million reads are unmapped? If so why don't they have the unmapped flag in the samfile.

[nsl]

Hi Andrea (aeveland) and anyone who has insight,

I have the same question about option -c in cuffdiff. Why are the defaults so different and how do they translate. If you found the answer would you mind sharing it with me.
I have been teaching myself rna-seq analysis for a little under 2 months and am using Galaxy to get my feet wet. The same default option -c there for cuffdiff is set at a 1000.
what is a conservative setting any suggestions?

Isthuthiy = thanks

nsl

[nsl]

Hi Cole,

I am using galaxy for my rna-seq analysis and am about to execute cuffdiff and am unsure about option -c. The default is a 1000 whereas the manual suggests 10. Could you please advice me?

Thanks

[ega2d]

Missing cuffcompare outputs

Hi all,

I have completed the cufflinks/cuffmerge/cuffdiff pipeline, but I realized that my cuffmerge isoforms.fpkm_tracking file did not contain class codes...I also noticed that the sensitivity/specificity analysis and other cuffcompare output files were not produced. Is there some way to make cuffmerge produce these files as well? Could I run cuffmerge with the -r and -R options for cuffcompare to have my tracking files include class codes?

I might be missing something major, but in any case...any help/pointers would be appreciated.

Thanks!