[wacguy]

-p flag causes problems

Hi,

I'm running tohat2 against the sequence of my transgene (luciferase). I only use the bowtie2 indexes and the FASTA file (i.e., no GTF).
I realized that using the -p 2 (or more) comparing to no -p option (but I didn't try to explicitly run w/ -p 1) can sometimes (this is file related because it doesn't haapen w/ all files) spits the following error:
Reporting output tracks
[FAILED]
Error running /usr/libexec/tophat.

Another type of error occured on some files:

Warning: Empty fasta file: '/rdata/ngseq/Playground/guy/tophat2/030314_root4/130814luc/P4/tmp/segment_juncs.fa'
Warning: All fasta inputs were empty
Error: Encountered internal Bowtie 2 exception (#1)
Command: /usr/bin/bowtie2-build /rdata/ngseq/Playground/guy/tophat2/030314_root4/130814luc/P4/tmp/segment_juncs.fa /rdata/ngseq/Playground/guy/tophat2/030314_root4/130814luc/P4/tmp/segment_juncs
[FAILED]
Error: Splice sequence indexing failed with err =1

I solved it by the flag: --no-novel-juncs
It is starnge because according to tohat manual, this flag is ignored w/o the -j/G option but it solved the problem.

topaht version 2.0.9 or 2.012.
Good luck,
Guy

[krespim]

same in v2.0.13

I am getting the same error (fail when reporting output tracks) on tophat 2.0.11. I have made three mapping runs for each of 12 samples. Of these, only the ones using both --no-discordant and --no-mixed together fail (I have not tried these two parameters individually), and runs with these options fail for all 12 samples.

My reads have been trimmed, and pairs may not be of equal size. But as far as I understand, tophat should be able to handle that. At least, it certainly seems to when not using these particular parameters. Number of paired reads are around 30-40 M.

Increasing RAM did not help in my case.

I had the same experience with v2.0.13. Tophat stopped reporting the error when the 2 options, --no-discordant and --no-mixed, where removed.

[ngoc_nguyen]

Error: [Errno 16] Device or resource busy

Hello, I'm new here.
I'm running tophat 2.0.10 with Bowtie 2.1.0.0 to map SE RNAseq to human genome. I have run the same tophat command with several datasets; some of my experiments exit fine with both accepted_hits.bam and unmapped.bam files produced and no tmp files left; some of the datasets fail without outputing unmapped.bam files, but tmp files still left in tophat output (I guess it is due to unfinished tophat run). For those runs that failed, it always reported error Error: [Errno 16] Device or resource busy, like this

[2016-05-12 15:34:26] Mapping left_kept_reads.m2g_um_seg1 to genome segment_juncs with Bowtie2 (1/2)
[2016-05-12 15:34:35] Mapping left_kept_reads.m2g_um_seg2 to genome segment_juncs with Bowtie2 (2/2)
[2016-05-12 15:34:43] Joining segment hits
[2016-05-12 15:39:47] Reporting output tracks
[FAILED]
Error: [Errno 16] Device or resource busy
Found 63037 junctions from happy spliced reads

[samopen] SAM header is present: 86 sequences.

Interestingly, the error seems to very consistently occur with some particular datasets, mean if a dataset fails with the fisrt tophat run, it keeps failing when I repeat running; but sometimes when I used the datasets that have successfully exit the tophat before, I have the tophat output failure in the next repeating run; so I don;t think it is because of the input file format problem.

And, if I converted the accepted_hits.bam files in tophat output to fastq files and remap them again using tophat, all of the run have exact this error no matter what the previous runs exit successfully or failed or not.

I had run tophat 1.4.1 with the same datasets before n did not get any errors.

I have search but havent found this kind of error " Error: [Errno 16] Device or resource busy" reported before. Any advice would be much appreciated!