question about mapping multiple samples

[Jacob Musser]

Hello,

After reading the manual and running a number of tests I am still having difficult mapping multiple samples with STAR. My interpretation of the manual is that you can map multiple samples in a single job, and receive output files for each individual sample. Is this correct or am I misinterpreting the manual?

Here is the command I have been using to run STAR on our cluster:

STAR --runThreadN 8 --quantMode GeneCounts --outMultimapperOrder Random --genomeDir /star_genome_index_gtf/ --sjdbGTFfile /star_genome_index_gtf/Trichoplax_scaffolds_JGI_AUGUSTUS_maxtrack2.gtf --sjdbGTFfeatureExon CDS --readFilesCommand gunzip -c --readFilesIn /picked_cell_fastq/000000000-BL84Y_10_RC2_18s000021-1-1_Varoqueaux_lane118s000021_1_sequence.txt.gz,/picked_cell_fastq/000000000-BL84Y_11_CC9_18s000022-1-1_Varoqueaux_lane118s000022_1_sequence.txt.gz /picked_cell_fastq/000000000-BL84Y_10_RC2_18s000021-1-1_Varoqueaux_lane118s000021_2_sequence.txt.gz,/picked_cell_fastq/000000000-BL84Y_11_CC9_18s000022-1-1_Varoqueaux_lane118s000022_2_sequence.txt.gz

As you can see, I have two paired-end samples, which I input using two comma-delimited lists that are separated by a space (sample1_reads1.fq,sample2_reads1.fq sample1_reads2.fq,sample2_reads2.fq). When I run this I get a single sam alignment file, and one ReadsPerGene file with a single column. So, it looks like STAR is treating my different samples as a single sample, and lumping the results together. Is there a way to produce output files for each sample?

Jake

[Alexander Dobin]

Hi Jake,

if you run multiple samples with one command, all the output will be written into the same files.

If you want separte outputs, you would have to write a loop over the input files, and start each run in a separate directory,

or add distinct --outFileNamePrefix to each run.

Alternatively, you can specify multiple RG tags which will be added to each read - then you can split the resulting BAM file into separate files by read group, e.g. --outSAMattrRGline --outSAMattrRGline ID:s1 , ID:s2 , ID:s3

Note that in this list the commas have to be separated by spaces.

Another convenience option implemented in 2.5.4a is --readFilesPrefix, where you can define a constant file prefix for all samples and thus simplify the --readFilesIn list.

Cheers

Alex