[alexdobin]

I am unable to get STAR use the SHARED MEMORY option between different instances. I followed the instructions given at

http://seqanswers.com/forums/showpost.php?p=107990&postcount=53

I am trying to run two instances (with delays of at least 2-3 mins), but each individual instance seems to allocate its own memory

$top
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
5291 xxx 20 0 26.6g 25g 25g R 100 20.6 17:37.64 STAR
5299 xxx 20 0 26.6g 25g 25g R 100 20.4 13:18.87 STAR

What else can I try?

It seems to me that the shared memory is working fine - since ipcs shows two instances attached to the same shared memory piece. What you see in 'top' is memory usage per process - and 25GB out of 26.6GB are shared for each process. To test it you can try to run >5 jobs at the same time - they would not be able to run without sharing memory.

[dpryan]

Can I use STAR's sam output to call SNP? If so, how?

Just search this site for "RNAseq SNP" for a plethora of examples, like this.

[ymc]

Can I use STAR's sam output to call SNP? If so, how?

[wildtypegoose]

I am unable to get STAR use the SHARED MEMORY option between different instances. I followed the instructions given at

http://seqanswers.com/forums/showpost.php?p=107990&postcount=53

I am trying to run two instances (with delays of at least 2-3 mins), but each individual instance seems to allocate its own memory

$top
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
5291 xxx 20 0 26.6g 25g 25g R 100 20.6 17:37.64 STAR
5299 xxx 20 0 26.6g 25g 25g R 100 20.4 13:18.87 STAR

What else can I try?

STAR Version: STAR_2.3.0e.Linux_x86_64

$ uname -a
Linux XXX 2.6.32-21-server #32-Ubuntu SMP Fri Apr 16 09:17:34 UTC 2010 x86_64 GNU/Linux


$ipcs
------ Shared Memory Segments --------
key shmid owner perms bytes nattch status
0x17000006 2588672 xxx 666 28271287966 2

------ Semaphore Arrays --------
key semid owner perms nsems

------ Message Queues --------
key msqid owner perms used-bytes messages

[alexdobin]

Thanks for your reply. I think I can count spanning chimeric junctions from Chimeric.out.junction. How can I count "encompassing"? Which output file should I look at?

The Chimeric.out.junction file contains the encompassing junctions as well. They are marked with -1 in column 7 (junction type). Of course, to assign the encompassing reads to a chimeric junction, you have to know the coordinates of the junction, or somehow cluster the inner ends of the encompassing mates.

[ymc]

STAR can detect chimeric alignments both "spanning" and "encompassing" chimeric junctions. However, you would need to do all the post-processing: filtering alignments, collapsing the chimeric junctions, annotating fused genes. There is some discussion about it on the STAR forum: https://groups.google.com/d/msg/rna-...U/yxj5C8LaovIJ

Thanks for your reply. I think I can count spanning chimeric junctions from Chimeric.out.junction. How can I count "encompassing"? Which output file should I look at?

[alexdobin]

Are there any plan to make STAR works for fusion genes? The existing tools are too slow for me...

STAR can detect chimeric alignments both "spanning" and "encompassing" chimeric junctions. However, you would need to do all the post-processing: filtering alignments, collapsing the chimeric junctions, annotating fused genes. There is some discussion about it on the STAR forum: https://groups.google.com/d/msg/rna-...U/yxj5C8LaovIJ

[ymc]

Are there any plan to make STAR works for fusion genes? The existing tools are too slow for me...

[kevinrue]

Examples of STAR usage in documentation?

Dear Alex and Shawn,

Could you please try to run 2-3 jobs (without killing your server ) pausing for 10sec between them, and send me the Log.out outputs for each job.

Our server is slightly busy at the moment (Nicolas Nalpas si pulling all the blanket for himself ), but I'll try asap.

It does make sense that jobs submitted simultaneously can't see the genome being loaded into shared memory (we have a looped script submitting a bunch of STAR jobs all within a few seconds from each other). For the record, we recently submitted 10 jobs in such a loop, and we went over our 256GB RAM + 4GB swap, which slowed the server down.

I think the documentation should include some examples because the explanation is a little confusing.

I support this idea, as we previously also had weird experiences with the outFilter(NminMatch) and (others) parameters which seem to filter on a number of consecutive matches rather than total number of matches for instance. I'd rather have detailed description (or maybe a short and a more detailed) of each option than a doubt which requires additional testing and guessing which impeges on my real project time.

Regarding the original point, example usage would be appreciated too. Maybe users could actually participate to such an effort, as we could rapidly gather a diversity of applications along with the combination of options we succesfully used? If so, we'd need some other place than this thread to share our commands.

Kevin

[Nicolas Nalpas]

Dear Alex,

I actually sorted it out by reading again your previous answer, so I tried out with --genomeSAindexNbases to 7 and it seems to work fine now, no error so far.

Thanks again for all your help, very appreciated.
Regards,
Nicolas

[Nicolas Nalpas]

Dear Alex,

Thanks for your help, I have tried to generate my Mycobacterium bovis genome as you rocommanded:

STAR --runMode genomeGenerate --genomeDir /workspace/storage/genomes/Mycobacterium_bovis/NC_002945.3/STAR2.3.0e --genomeFastaFiles /workspace/storage/genomes/Mycobacterium_bovis/NC_002945.3/source_file/Mycobacterium_bovis_NC_002945.3.fasta --genomeSAindexNbases 8 --runThreadN 1

And I now obtain a different error while doing the alignment:

Jun 20 09:23:01 ...... FATAL ERROR, exiting
Jun 20 09:23:01 ..... Started STAR run

EXITING because of FATAL error, could not open file /path/sjdbInfo.txt
SOLUTION: check that the path to genome files, specified in --genomDir is correct and the files are present, and have user read permsissions

So I checked the genome generate directory, into which read and write permissions are allowed, however there is no such "sjdbInfo.txt" file.

My command for alignment was:

STAR --runMode alignReads --genomeDir /path/STAR2.3.0e --genomeLoad LoadAndKeep --readFilesIn /path/N1178_CN_0H_pe1.fastq /path/N1178_CN_0H_pe2.fastq --runThreadN 1 --outFilterMultimapNmax 10 --outSAMmode Full --outSAMattributes Standard --outFileNamePrefix ./N1178_CN_0H_ --outReadsUnmapped Fastx

Any idea on how I can sort this out?

Thanks a lot for all your help.
Regards,
Nicolas

[alexdobin]

Hi everyone,

At the moment, I am having an issue with STAR 2.3.0e, which I am trying to use in order to align 2x90bases paired-end reads sample against Mycobacterium bovis genome NC_002945.3 from NCBI.
I obtain the error below:

Jun 19 19:45:09 ..... Started STAR run
Jun 19 19:45:09 ..... Started mapping
Segmentation fault (core dumped)

Nicolas

Hi Nicolas,

This is a known problem for very small genomes. At the genome generation step, please try to reduce the value of --genomeSAindexNbases to <=8, and then re-run the mapping step.
Generally, --genomeSAindexNbases needs to be scaled with the genome length, as ~min(14,log2(ReferenceLength)/2 - 1). I will need to incorporate this scaling in the future releases.

Cheers
Alex

[alexdobin]

Hi Kevin, Shawn,

it's great that Shawn's work-around worked. However, the --genomeLoad LoadAndRemove option is supposed to work the same way, allowing one copy of the genome to be shared between the jobs. The only thing I can think of at the moment is that if you submit the jobs precisely at the same moment, they might not see each other shared memory, and decide to allocate it on its own. Could you please try to run 2-3 jobs (without killing your server ) pausing for 10sec between them, and send me the Log.out outputs for each job. Also, while they are running, you can run Linux 'ipcs' command which will tell us which shared memory pieces are being used.

The fact that 50GB of RAM is used for the 27GB index also concerns me a bit.

Cheers
Alex

[Nicolas Nalpas]

Hi everyone,

At the moment, I am having an issue with STAR 2.3.0e, which I am trying to use in order to align 2x90bases paired-end reads sample against Mycobacterium bovis genome NC_002945.3 from NCBI.
I obtain the error below:

Jun 19 19:45:09 ..... Started STAR run
Jun 19 19:45:09 ..... Started mapping
Segmentation fault (core dumped)

The command used to generate genome was (which seems to be working fine):

STAR --runMode genomeGenerate --genomeDir /path/STAR2.3.0e --genomeFastaFiles /path/Mycobacterium_bovis_NC_002945.3.fasta --runThreadN 1

The command used to align reads was (which gives me the error above):

STAR --runMode alignReads --genomeDir /workspace/storage/genomes/Mycobacterium_bovis/NC_002945.3/STAR2.3.0e --genomeLoad NoSharedMemory --readFilesIn /home/dmagee/scratch/ALV_MAC_RNAseq/fastq_sequence/N1178_CN_0H_pe1.fastq /home/dmagee/scratch/ALV_MAC_RNAseq/fastq_sequence/N1178_CN_0H_pe2.fastq --runThreadN 1 --outFilterMultimapNmax 10 --outSAMmode Full --outSAMattributes Standard --outFileNamePrefix ./N1178_CN_0H_ --outReadsUnmapped Fastx

I am not sure of what is going on there, since my command was working perfectly when I aligned the exact same reads sample against Bos taurus genome.

Any idea on what could be the problem, it will be very much appreciated.
Thanks a lot.
Best wishes,
Nicolas

[sdriscoll]

I think the documentation should include some examples because the explanation is a little confusing. Glad that worked, though.