[Xi Wang]

Hi Cole,

does Cufflinks take as input the 'raw' SAM file produced by Bowtie (>version 0.11.1)?

Daniel

From Tophat manual:
The SAM file supplied to Cufflinks must be sorted by reference position. If you aligned your reads with TopHat, your alignments will be properly sorted already. If you used another tool, you may want to make sure they are properly sorted as follows:

sort -k 3,3 -k 4,4n hits.sam > hits.sam.sorted

And you may also have to filter the reads not mapped to the reference genome, marked by "*" in the field of RNAME.

[GoneSouth]

Some problems with cufflinks

Hi guys,

I have now tried it several times but I cant seem to get cufflinks running with my new 3x paired end lanes, I always get a 'bus error'

since I always get the error at slightly different positions
Processing bundle [ 3:9131617-9139931 ] with 8255 non-redundant alignments
zsh: bus error cufflinks -m 85 -I 50000 -p 4 -L VPG

Processing bundle [ 3:9411509-9430092 ] with 11575 non-redundant alignments
Bus error

I suspect a multithreading error. Has anyone encountered a similar problem or even better knows how to deal with this problem?

Furthermore I frequently receive strange warnings:
Warning: restimation failed, importance samples have zero weight
Warning: ITERMAX reached in abundance estimation, estimation hasn't fully converged
Warning: Fisher matrix is not positive definite (bad element: -2.99551e-08)

Does anyone know if they are critical or what do they mean?


To provide some additional information for the bus error. I have already successfully run cufflinks on single lane paired end data. but when I merge 3 lanes of paired end data the 'bus error' occurs. I mapped the reads with TopHat and merging was done with the following command
cat s1/sample1.sam s2/sample2.sam s3/sample3.sam| sort -k 3,3 -k 4,4n >all/all.sam

I would greatly appreciate any comments and or help on this issues!

all the best ro

[William G. Spollen]

boost 1.42.0 and cufflinks 0.8.1 compatability?

When cufflinks version 0.8.1 was released on Feb. 13, 2010, had it been tested using the boost Version 1.42.0, which was released on Feb. 2, 2010? I am having trouble getting cufflinks to compile because of some problem in finding all that it needs in boost, and the file structure of boost had changed a bit with the new version. So far using CFLAGS and the appropriate path to the boost/property_map includes is not helping and I may just try an earlier version of boost. Has anybody had this problem?

Thanks
Bill

[siiner]

cufflinks mem consumption problem

Hi Cole,

Is there any memory controlling methods in the newly released version 0.8.2?
I've got a big memory problem when processing an RNA-Seq mapping result, which contains a really huge bundle with more than 2,000,000 reads. Even a server with 32GB mem could not finish this.

Can you help me ?

Regards,

Bob

[hrajasim]

gene ID and transcript ID

I have completed the cufflinks, cuffcompare and cuffdiff runs and have the results. However, how do I go from the cufflinks IDs to the reference gene symbol? I did not give the gff file as input when I ran. Is there any way other than re-running everything with gff file?

Thanks,

[Xi Wang]

I have completed the cufflinks, cuffcompare and cuffdiff runs and have the results. However, how do I go from the cufflinks IDs to the reference gene symbol? I did not give the gff file as input when I ran. Is there any way other than re-running everything with gff file?

Thanks,

I think you'd better rerun Cuffcompare with the "-r" option, and then run cuffdiff again. Maybe this is the easiest way.

[lahoman]

discovery of novel transcript

Hi, there,

I'm tring to find the novel transcript using the following way.

1. /tophat-1.2.0.Linux_x86_64/tophat -r 150 Human_bowtie_index reads.000.1.fastq reads.000.2.fastq --GTF Homo_sapiens.GRCh37.60.gtf

2. /tophat-1.2.0.Linux_x86_64/tophat -r 150 -o nonovel Human_bowtie_index reads.000.1.fastq reads.000.2.fastq --GTF Homo_sapiens.GRCh37.60.gtf --no-novel-juncs


Then I'm going to compare two junction.bed files to see the difference.
Is it correct? How do you think?

Thanks a lot,

Lahoman