Skip to content

STAR

Description

STAR (Spliced Transcripts Alignment to a Reference) is a fast NGS read aligner for RNA-seq data.

License

Free to use and open source under GNU GPLv3.

Available

Version on CSC's Servers

Puhti: 2.7.10a

Usage

The STAR commands listed abowe are activated by loading biokit module.

module load biokit

Before you can run the actual alignment job, you must index your fasta formatted reference genome. In Puhti the working copies of reference genome indexes, as well as any large files, should be stored to the scatch directory ($SCRATCH). In this example we store the indexes to the directory $SCRATCH/star-genomes.

First the reference genome index directory is generated with command: 

mkdir $SCRATCH/star-genome
After that, the indexing can be done with command: 
STAR --runMode genomeGenerate --genomeDir $SCRATCH/star-genome --genomeFastaFiles /path/to/genome/genome.fasta --runThreadN 2
Once the indexing is done the actual mapping task can be launched. STAR will generate the mapping output using fixed file names. Because of that it is recommended that each STAR job is run in a new, empty directory. In Puhti you should create this new job directory to scratch directory ($SCRATCH) of your project. New directory called starjob1 can be created with command: 
mkdir $SCRATCH/starjob1
after that the actual mapping job can be launched with commands: 
cd $SCRATCH/starjob1
STAR --genomeDir $SCRATCH/star-genomes --readFilesIn my_reads.fastq
The default parameters, that STAR uses, are typical for mapping 2x76 or 2x101 Illumina reads to the human genome. In Puhti the default parameter settings are available in file: 
/appl/soft/bio/star/STAR-2.7.2a/source/parametersDefault
In Puhti, all heavier computing tasks should be executed as batch jobs. In batch jobs you can also utilize thread based parallelization. Bellow is a sample batch job file for STAR. The job uses six computing cores from a single computing node. The memory reservation is 24 GB (4GB/core * 6 cores). Note that you must change the --account setting to match you poject. 
#!/bin/bash -l
#SBATCH --job-name=STAR
#SBATCH --output=STAR.stdout
#SBATCH --error=STAR.stderr
#SBATCH --partition=small
#SBATCH --ntasks=1
#SBATCH --nodes=1
#SBATCH --cpus-per-task=6
#SBATCH --account=project_1234567
#SBATCH --mem=24000

# calculate indexes. You don't need to recalculte the indexes if they already exist.
mkdir $SCRATCH/star-genome
STAR --runMode genomeGenerate --genomeDir $SCRATSCH/star-genome --genomeFastaFiles /path/to/genome/genome.fasta --runThreadN $SLURM_CPUS_PER_TASK

# Run the mapping task
STAR --genomeDir $SCRATCH/star-genome --readFilesIn my-reads.fastq --runThreadN $SLURM_CPUS_PER_TASK

The batch job script is launced with command sbatch. For example: 

sbatch starjob1.sh

Manual