rsem-calculate-expression fails to open iso_res file

Question

[emyers]

I am new to any kind of sequence analysis, and am trying to use RSEM for mapping (with STAR) and expression estimates. The step of preparing a reference completed successfully. The call to rsem-calculate-expression gets through the mapping, but hits an error before producing the expression estimates.

This is the call to rsem-calculate-expression:

sudo /opt/RSEM-1.3.0/rsem-calculate-expression $trimmedFile $refName BF_RORbHTp200_1 -p 8 --star --star-path $starPath --star-output-genome-bam

And the output:

/opt/STAR/bin/MacOSX_x86_64//STAR --genomeDir /Volumes/CodingClub1/STAR_stuff/rsem_reference  --outSAMunmapped Within  --outFilterType BySJout  --outSAMattributes NH HI AS NM MD  --outFilterMultimapNmax 20  --outFilterMismatchNmax 999  --outFilterMismatchNoverLmax 0.04  --alignIntronMin 20  --alignIntronMax 1000000  --alignMatesGapMax 1000000  --alignSJoverhangMin 8  --alignSJDBoverhangMin 1  --sjdbScore 1  --runThreadN 8  --genomeLoad NoSharedMemory  --outSAMtype BAM Unsorted  --quantMode TranscriptomeSAM  --outSAMheaderHD \@HD VN:1.4 SO:unsorted  --outFileNamePrefix BF_RORbHTp200_1.temp/BF_RORbHTp200_1  --readFilesIn /Volumes/CodingClub1/RNAseq/RORb/trimmed/trimmed_m20_q20/BF_RORbHTp200_1_trimmed_m20_q20.fastq 
Dec 18 09:36:35 ..... started STAR run
Dec 18 09:36:35 ..... loading genome
Dec 18 09:39:16 ..... started mapping
Dec 18 09:43:09 ..... finished successfully

rsem-parse-alignments /Volumes/CodingClub1/STAR_stuff/rsem_reference/mm10_ucsc_ BF_RORbHTp200_1.temp/BF_RORbHTp200_1 BF_RORbHTp200_1.stat/BF_RORbHTp200_1 BF_RORbHTp200_1.temp/BF_RORbHTp200_1.bam 1 -tag XM

rsem-build-read-index 32 1 0 BF_RORbHTp200_1.temp/BF_RORbHTp200_1_alignable.fq

rsem-run-em /Volumes/CodingClub1/STAR_stuff/rsem_reference/mm10_ucsc_ 1 BF_RORbHTp200_1 BF_RORbHTp200_1.temp/BF_RORbHTp200_1 BF_RORbHTp200_1.stat/BF_RORbHTp200_1 -p 8 -b BF_RORbHTp200_1.temp/BF_RORbHTp200_1.bam 0

Fail to open file BF_RORbHTp200_1.temp/BF_RORbHTp200_1.iso_res!

RSEM appears to be writing other files to the directory BF_RORbHTp200_1.temp just fine. The other possibility that occured to me is the number of files that can be opened at the same time, in case RSEM was opening a lot of temporary files. I set ulimit -n to 2048 and got the same problem. Any help would be greatly appreciated.

[mforde84]

sounds like a permissions issue. why are you running as root?

[emyers]

Whoops - I shouldn't have been. But I tried again under a regular user and got the same error.

[mforde84]

look at the permissions of the folder in question. check that it's readable, and writable to the user running the process 'ls -l'. check that you have available space on the mount point 'df -h'.

if all else fails, delete the temporary folders and files associated with the RSEM analysis, and rerun the analysis from scratch from a normal user.

[emyers]

Thanks very much for your suggestions. The directory is read/writeable to the user, and I tried again on a drive with terabytes of available space, starting from scratch, from a regular user. I'm getting the same strange error.

I have been able to do the mapping part with STAR entirely separately from RSEM. . .maybe for now I should use another tool for getting expression estimates, like Cufflinks.

I hate to admit defeat, so if you do think of any other possibilities I'd love to hear it! Thanks again for your help.