FAQ

General

What is fq-compressor?

fq-compressor is a high-performance FASTQ compression tool that uses domain-specific algorithms (Assembly-based Compression for sequences, Statistical Context Mixing for quality scores) to achieve near-theoretical-limit compression ratios while supporting random access and parallel processing.

How does it compare to gzip/zstd?

General-purpose compressors treat FASTQ data as arbitrary text. fq-compressor exploits the biological structure of sequencing data to achieve 3-5x better compression ratios. See the Benchmarks page for detailed comparisons.

Is it lossless?

Yes. fq-compressor performs fully lossless compression by default. Every base, quality score, and identifier is preserved exactly.

Compatibility

What FASTQ formats are supported?

• Standard 4-line FASTQ (.fastq, .fq)
• Gzip-compressed FASTQ (.fastq.gz, .fq.gz)
• Bzip2-compressed FASTQ (.fastq.bz2, .fq.bz2)
• XZ-compressed FASTQ (.fastq.xz, .fq.xz)
• Single-end (SE) and paired-end (PE) data
• Illumina, BGI, and other common identifier formats

Zstandard-compressed FASTQ input (.zst) is not supported yet.

What read lengths are supported?

fq-compressor handles all read lengths, but uses different strategies:

Can I use .fqc files with other tools?

No — .fqc is a custom format. You need to decompress with fqc decompress before using standard bioinformatics tools. The random access feature makes it easy to extract subsets without full decompression.

Performance

How many threads should I use?

For best results, use the number of physical CPU cores. The --pipeline flag is recommended for files larger than 100 MB.

fqc compress -i reads.fastq -o reads.fqc -t 8 --pipeline

How much memory does it need?

Memory usage depends on the block size and number of in-flight blocks. The default memory limit is 8 GB, adjustable via --memory-limit:

fqc compress -i reads.fastq -o reads.fqc --memory-limit 4096  # 4 GB

Why is compression slower than decompression?

Compression involves computationally expensive operations (minimizer extraction, read reordering, consensus generation) that don't have equivalents during decompression. This asymmetry is by design — compress once, decompress many times.

Troubleshooting

"Format error" when decompressing

Ensure the file was created by fq-compressor and hasn't been corrupted:

fqc verify archive.fqc

Out of memory during compression

Reduce the memory limit to force smaller block sizes:

fqc compress -i reads.fastq -o reads.fqc --memory-limit 2048

Build fails with "TBB not found"

Make sure Conan installed dependencies correctly:

conan install . --build=missing -of=build

Academic

How should I cite fq-compressor?

If you use fq-compressor in your research, please cite the GitHub repository:

LessUp. fq-compressor: High-performance FASTQ compression with random access.
https://github.com/LessUp/fq-compressor

What algorithms does it build on?

• Spring (Chandak et al., 2019) — Assembly-based sequence compression
• Mincom (Liu et al.) — Minimizer bucketing
• fqzcomp5 (Bonfield) — Quality score context mixing
• HARC (Chandak et al., 2018) — Haplotype-aware reordering