Starlong

Common Patterns

# Build a genome index
STARlong \
  --runMode genomeGenerate \
  --genomeDir starlong_index \
  --genomeFastaFiles genome.fa \
  --sjdbGTFfile genes.gtf \
  --runThreadN 16

# Align long reads
STARlong \
  --genomeDir starlong_index \
  --readFilesIn longreads.fastq \
  --runThreadN 16

# Align gzipped reads and emit coordinate-sorted BAM
STARlong \
  --genomeDir starlong_index \
  --readFilesIn longreads.fastq.gz \
  --readFilesCommand zcat \
  --outSAMtype BAM SortedByCoordinate \
  --runThreadN 16

Recommended Workflow

1. Generate a compatible STAR genome index, optionally with --sjdbGTFfile for splice annotations.
2. Run the STARlong wrapper rather than hard-coding one SIMD binary unless you specifically need a fixed backend.
3. Add --readFilesCommand zcat or a similar decompressor for compressed inputs.
4. Review the standard STAR outputs along with long-read-specific alignment quality and splice-junction summaries.

Guardrails

• STARlong is a wrapper script, not the aligner binary itself. bash -x STARlong --version on this host showed it selecting STARlong-avx2.
• --help shows the generic STAR usage banner (Usage: STAR ...) because the wrapper forwards directly to the selected backend.
• --version works and returned 2.7.11b locally.
• Help text declares versionGenome 2.7.4a as the earliest compatible genome index version for this release.
• Genome FASTA files for --runMode genomeGenerate must be plain text and cannot be zipped.
• Default output is SAM (outSAMtype SAM), so request BAM explicitly when needed.
• Help text exposes STARlong-specific window coverage controls such as winReadCoverageRelativeMin and winReadCoverageBasesMin; adjust them only if you intentionally tune long-read sensitivity.