STAR is a parameter-rich RNAseq aligner. This tool is based on this release: https://github.com/alexdobin/STAR/tree/2.7.10a. Depending on the parameters used, run time is typically around 5-10 hours, with spot instance cost of $2.50 - $8.00
tools/star_2.7.10a_align.cwl
genomeDir
: Reference generated using Reference generation tool. Will be tar gzipped, and tool will unpack at run time.
readFilesIn1
: Fastq reads file, either uncompressed or gzippedreadFilesIn2
: Required IF data is paired end. Same format asreadFilesIn1
outSAMattrRGline
: Suggested setting, with TABS SEPARATING THE TAGS, format is:ID:sample_name LB:aliquot_id PL:platform SM:BSID
. For example ID:
7316-242 LB:750189 PL:ILLUMINA SM:BS_W72364MN
outFileNamePrefix
: output files name prefix (including full or relative path). Can only be defined on the command line. Tool will add '.' after prefix to easily delineate between file name and suffix
Try not to panic, but there are many options. We do set many defaults to make life easier, and should yield reasonable results. Below is a table of params and values we set by default, as well as suggested changes based on downstream fusion callers, which should have little effect on expression calculations, but potentially valuable increases in sensitivity for fusion calling.
Param/Description | WF Default (arriba recommended) | GTEx/Broad | STAR-Fusion Recommended |
---|---|---|---|
runThreadN | 16 | 16 | 16 |
twopassMode | Basic | Basic | Basic |
alignSJoverhangMin | 8 | 8 | 8 |
outFilterMismatchNoverLmax | 0.1 | 0.1 | 0.1 |
outFilterType | BySJout | BySJout | BySJout |
outFilterScoreMinOverLread | 0.33 | 0.33 | 0.33 |
outFilterMatchNminOverLread | 0.33 | 0.33 | 0.33 |
outReadsUnmapped | None | None | None |
limitSjdbInsertNsj | 1200000 | 1200000 | 1200000 |
outSAMstrandField | intronMotif | intronMotif | intronMotif |
outFilterIntronMotifs | None | None | None |
alignSoftClipAtReferenceEnds | Yes | Yes | Yes |
quantMode | TranscriptomeSAM GeneCounts | TranscriptomeSAM GeneCounts | TranscriptomeSAM GeneCounts |
outSAMtype | BAM Unsorted | BAM Unsorted | BAM Unsorted |
outSAMunmapped | Within | Within | Within |
genomeLoad | NoSharedMemory | NoSharedMemory | NoSharedMemory |
chimMainSegmentMultNmax | 1 | 1 | 1 |
outSAMattributes | NH HI AS nM NM ch | NH HI AS nM NM ch | NH HI AS nM NM ch |
alignInsertionFlush | None | None | Right |
alignIntronMax | 1000000 | 1000000 | 100000 |
alignMatesGapMax | 1000000 | 1000000 | 100000 |
alignSJDBoverhangMin | 1 | 1 | 10 |
outFilterMismatchNmax | 999 | 999 | 999 |
alignSJstitchMismatchNmax | 5 -1 5 5 | 0 -1 0 0 | 5 -1 5 5 |
alignSplicedMateMapLmin | 0 | 0 | 30 |
alignSplicedMateMapLminOverLmate | 0.5 | 0.66 | 0 |
chimJunctionOverhangMin | 10 | 15 | 8 |
chimMultimapNmax | 50 | 0 | 20 |
chimMultimapScoreRange | 1 | 1 | 3 |
chimNonchimScoreDropMin | 20 | 20 | 10 |
chimOutJunctionFormat | 1 | 1 | 1 |
chimOutType | Junctions WithinBAM SoftClip | Junctions SeparateSAMold WithinBAM SoftClip | Junctions WithinBAM SoftClip |
chimScoreDropMax | 30 | 20 | 20 |
chimScoreJunctionNonGTAG | -1 | -1 | -4 |
chimScoreSeparation | 1 | 10 | 10 |
chimSegmentMin | 10 | 15 | 12 |
chimSegmentReadGapMax | 3 | 0 | 0 |
outFilterMultimapNmax | 50 | 20 | 20 |
peOverlapMMp | 0.01 | 0.01 | 0 |
peOverlapNbasesMin | 10 | 0 | 12 |
log_progress_out
: Simple progress output. Can use to gauge speed and run timelog_out
: Contains a summary of all params used and reference fileslog_final_out
: Overall summary of read mapping statisticsgenomic_bam_out
: UNSORTED read mapping to genomic coordinatesjunctions_out
: high confidence collapsed splice junctions in tab-delimited formtranscriptome_bam_out
: Read mapping to transcriptomechimeric_sam_out
: Deprecated output. Incompatible with certain options, has chimeric read alignmentschimeric_junctions
: Chimeric junctions output file. May be used for downstream tools for fusion analysisgene_counts
: STAR-generated read counts by gene
tools/star_2.7.10a_genome_generate.cwl
This tool is used to create the necessary genome indices for STAR.
You should only need to do this once for each gene model, fasta reference, and possibly read length of input files (STAR author says values of 100 should work for most cases)
genomeDir
: Output dirname. Recommend STAR_{version}_GENCODE{version num}genome_fa
: Fasta file to index. Recommend from GENCODE, PRI assembly. Must unzip first if compressedgtf
: Matched GTF file to index. Recommend from GENCODE, PRI assembly
runThreadN
: Number of threads to use. Default 16sjdbOverhang
: Splice junction database overhang. Ideal value is read len minus 1, but default 100 ok for most cases
star_ref
: A tar gzipped package with all required files and indices to run STAR