Error For End Coordinate When Running Cuffdiff

Question

[camelbbs]

Hi, while i run cuffdiff I got an error like this:

cuffdiff  -p 8  -o /s1_s2  -b  /bowtie2index/hg19.fa  -u  /all_merged/merged.gtf  /s1.bam  /s2.bam

[20:01:09] Loading reference annotation and sequence.
Error (GFaSeqGet): end coordinate (108996652) cannot be larger than sequence length 107349540
Error (GFaSeqGet): end coordinate (108996652) cannot be larger than sequence length 107349540
Error (GFaSeqGet): end coordinate (108996652) cannot be larger than sequence length 107349540
Error (GFaSeqGet): end coordinate (108996652) cannot be larger than sequence length 107349540
Error (GFaSeqGet): end coordinate (108996652) cannot be larger than sequence length 107349540
Error (GFaSeqGet): end coordinate (108996652) cannot be larger than sequence length 107349540
Error (GFaSeqGet): end coordinate (108996652) cannot be larger than sequence length 107349540
Error (GFaSeqGet): end coordinate (108996652) cannot be larger than sequence length 107349540

Anyone knows this? Thanks!!

[Arun]

Are you running on a 32-bit system?

[camelbbs]

running on unix cluster. Do you mean that is because of cpu or memory exhausted?

[Arun]

Then most likely not. I was just wondering if the allocation was not possible due to memory restrictions of a 32-bit architecture.

[Picasa]

I have the same error even without using the -s in cuffmerge ... any solution ?

Answer 1

[matted]

It's complaining about a read or annotation having a position larger than the reference chromosome length. 107349540 is the length of chromosome 14 in hg19, so I'd guess maybe you're using mouse reads mistakenly with a human reference here (mouse chromosome 14 is bigger than human: 125194864 bp). Basically, make sure the reference you hand to cuffdiff is the same as was used to generate the BAMs.

[camelbbs]

I am sure I am using all hg19 reference to tophat and cuffdiff, may be the problem is the parameter -u?

Answer 2

[Arun]

There seems to be a similar post and an explanation here (suggested solution seems to be to run cufflinks with out the -b option).

[camelbbs]

Thanks so much. While I run cuffdiff without -b option, it seems work. So which one is better to run with -b option, cufflinks or cuffdiff?

Answer 3

[pengchy]

I also encounter this problem. I check the output transcripts.gtf, there indeed the coordinate larger than the reference genome sequence length. However, when I rerun cufflinks for the error scaffolds specifically, the output transcripts.gtf become normal. Both running with -b and -u parameters.

Answer 4

[fta.mirzadeh]

I was getting this error using cuffmerge, ran cuffmerge without -s genome.fa option and it worked! dont know if was accurate or not but the program gave a merged.gtf as output. I proceeded to cuffdiff using THAT merged.gtf and received the error again! I omitted the -b and its running so far :)

Answer 5

[super]

Hi I have same problem with you! It seems that lots of people have this error.

I use the Ensembl genome(*dna.toplevel.fa) and GTF file.

If I ran cufflinks (even without -b genome.fa) but then run Cuffmerge with -s genome.fa, I got the error!

If I omitted -b genome.fa in both Cufflinks and Cuffdiff and meanwhile omit -s genome.fa in Cuffmerge, I didn't get the error! However, I don't know if it was accurate.