Computing time evaluate

[cindylanzao]

Hi,all

I am doing RNA-seq alignment with STAR, my workstation is dual core(E5-2620v3), 12 thread in total, 64G RAM. I am running 9 mouse RNA-seq sample data(1GB/sample), parallelly, two days has gone, still no result. No error, the log just say start mapping. Can any one with similar experience tell me how can I make it faster or is there anything wrong.

Below is my command:
/STAR-STAR_2.4.2a/bin/Linux_x86_64/STAR --outFilterIntronMotifs RemoveNoncanonicalUnannotated --runThreadN 2 --outTmpDir /pathdir/ --outSAMtype BAM SortedByCoordinate --genomeDir /pathfordatabaseformus/ --readFilesIn XXX.clean.fq.gz --readFilesCommand zcat

Thanks for all the guys who offer me help!

[alexdobin]

Originally Posted by cindylanzao

Hi,all

I am doing RNA-seq alignment with STAR, my workstation is dual core(E5-2620v3), 12 thread in total, 64G RAM. I am running 9 mouse RNA-seq sample data(1GB/sample), parallelly, two days has gone, still no result. No error, the log just say start mapping. Can any one with similar experience tell me how can I make it faster or is there anything wrong.

Below is my command:
/STAR-STAR_2.4.2a/bin/Linux_x86_64/STAR --outFilterIntronMotifs RemoveNoncanonicalUnannotated --runThreadN 2 --outTmpDir /pathdir/ --outSAMtype BAM SortedByCoordinate --genomeDir /pathfordatabaseformus/ --readFilesIn XXX.clean.fq.gz --readFilesCommand zcat

Thanks for all the guys who offer me help!

Hi @cindylanzao

you can check check Log.progress.out for the current mapping speed and progress.
Are you trying to map all 9 samples at the same time with 2 threads for each job? If so, you need to use shared memory option, e.g. --genomeLoad LoadAndKeep. Without shared memory, the genomes for each sample have to be loaded in RAM, which requires ~9*25 GB.
A simpler option would be to map one sample at a time with 12 threads.

Cheers
Alex

[cindylanzao]

Originally Posted by alexdobin

Hi @cindylanzao

you can check check Log.progress.out for the current mapping speed and progress.
Are you trying to map all 9 samples at the same time with 2 threads for each job? If so, you need to use shared memory option, e.g. --genomeLoad LoadAndKeep. Without shared memory, the genomes for each sample have to be loaded in RAM, which requires ~9*25 GB.
A simpler option would be to map one sample at a time with 12 threads.

Cheers
Alex

Thanks for your kind reply Alex, I have checked the Log.progress.out, it contains nothing but the title after a whole night's running with only one sample with 12 thread. Do you have any other suggestion? Thanks!

[alexdobin]

Originally Posted by cindylanzao

Thanks for your kind reply Alex, I have checked the Log.progress.out, it contains nothing but the title after a whole night's running with only one sample with 12 thread. Do you have any other suggestion? Thanks!

Hi @cindylanzao,

please post or email me the Log.out file from the failed single sample 12-thread run.

Cheers
Alex

[cindylanzao]

Originally Posted by alexdobin

Hi @cindylanzao,

please post or email me the Log.out file from the failed single sample 12-thread run.

Cheers
Alex

Thanks for your help, but the project is in a hurry, so I have decided to change to tophat as the aligner, and have deleted all the related output file from STAR.