mock-fastq-generator

mock-fastq-generator is an open-source software utility for generating targeted synthetic FASTQ datasets with controllable Phred quality score profiles. Available both as an installable Python package/CLI and as a standalone client-side web application, it allows bioinformatics developers to benchmark sequence processing tools (trimmers, aligners, denoisers) without relying on heavy empirical reference-genome simulators or exposing proprietary data.

---

Key Features

• Parametric Quality Decay Models: Choose between three distinct decay models:
  • Gaussian: Bounded bell-curve quality peak centered at a designated position.
  • Exponential: Simulates 3' end quality degradation with a sub-exponential initial plateau.
  • Sigmoidal: Sharp quality score drop centered at a specific nucleotide coordinate.
• Illumina 4-State Quality Binning (--binned_quality): Snaps Phred scores to discrete bins (Q2, Q12, Q23, Q37) emulating modern Illumina NovaSeq platforms.
• Homopolymer Quality Penalties (--homopolymer_penalty): Applies context-dependent $-10$ Phred score penalties inside runs of $>4$ identical consecutive nucleotides.
• Targeted Construct Assembly: Combines custom FASTA templates with adapter/flanking upstream sequences and randomized nucleotide margins.
• Paired-End Read Simulation (--paired-end): Generates matching R1 (forward) and R2 (reverse complement) paired FASTQ files with matching headers (/1 and /2).
• Inline Gzip Compression (--gzip): Directly writes .fastq.gz archives.
• Standard Output Streaming (--stdout): Streams FASTQ records directly to stdout for UNIX piping (|) into tools like bwa, fastp, or fastqc.
• JSON Configuration Profiles (--parameters_file): Load reproducible simulation settings from structured JSON files.
• Standalone Web Application: Complete client-side web interface (public/index.html) requiring no backend server.

---

Figure 1: Comparison of synthetic Phred quality score profiles generated across a 500-base construct. (A) Parametric Gaussian decay. (B) Exponential 3' quality dropoff. (C) Sigmoidal drop combined with 4-state NovaSeq binning and homopolymer penalties.

---

Installation

Prerequisites

• Python 3.8+
• numpy >= 1.20

Package Installation

Install directly from GitHub via pip:

git clone https://github.com/florez-alberto/mock-fastq-generator.git
cd mock-fastq-generator
pip install .

For active development (includes pytest and ruff):

pip install -e ".[dev]"

---

Usage

Command-Line Interface (CLI)

1. Basic Single-End Generation

mock-fastq-generator \
  --template_sequence template_example.fasta \
  --output_file synthetic.fastq

2. Advanced Paired-End with NovaSeq Binning & Gzip Compression

mock-fastq-generator \
  --template_sequence template_example.fasta \
  --output_file sample_data \
  --paired-end \
  --gzip \
  --decay_model sigmoidal \
  --decay_rate 0.05 \
  --center 100 \
  --binned_quality \
  --homopolymer_penalty \
  --number_of_sequences 1000

(Produces sample_data_R1.fastq.gz and sample_data_R2.fastq.gz)

3. UNIX Piping (Streaming directly into aligner or trimmer)

mock-fastq-generator \
  --template_sequence template_example.fasta \
  --number_of_sequences 500 \
  --stdout | bwa mem reference.fasta - > mapped_reads.sam

4. JSON Configuration File

Create a config.json file:

{
  "upstream_sequence": "GCCGGCCATGGCG",
  "left_margin": 15,
  "total_length": 300,
  "number_of_sequences": 200,
  "decay_model": "exponential",
  "decay_rate": 0.001,
  "min_val": 7,
  "max_val": 40,
  "score_type": "phred"
}

And execute:

mock-fastq-generator --template_sequence template_example.fasta --parameters_file config.json --output_file output.fastq

---

Python API

You can also import mock-fastq-generator into Python scripts or automated test suites:

from mock_fastq_generator import assemble_sequences, assemble_paired_sequences
from mock_fastq_generator.io import write_fastq

# Single-end records
records = assemble_sequences(
    template_sequence="GAATTCGCAAGCTT",
    upstream_sequence="GCCGGCCATGGCG",
    left_margin=15,
    total_length=150,
    number_of_sequences=100,
    decay_model="exponential",
    decay_rate=0.0008,
    min_val=7,
    max_val=40,
    score_type="phred",
    binned_quality=True
)

# Write to file
write_fastq("output.fastq.gz", records, compress=True)

---

Standalone Web Application

mock-fastq-generator includes a zero-dependency, client-side web application located in public/index.html.

• Open public/index.html directly in any modern web browser.
• Interactive controls allow configuring all simulation parameters.
• Real-time visualization of positional Phred score decay curves powered by Chart.js.
• Generates and downloads FASTQ files client-side using browser Blob APIs without server computation.

---

CLI Reference

Argument	Default	Description
--output_file	None	Output file path (or prefix for paired-end mode). Mutually exclusive with --stdout.
--stdout	False	Stream FASTQ content directly to stdout. Mutually exclusive with --output_file.
--template_sequence	None	Path to FASTA template file. Defaults to bundled template_example.fasta.
--parameters_file	None	Path to JSON configuration file overriding defaults.
--paired-end	False	Generate R1 forward and R2 reverse complement read pairs.
--gzip	False	Write gzip-compressed .fastq.gz output files.
--upstream_sequence	"GCCGGCCATGGCG"	Upstream adapter/flanking sequence prepended to reads.
--left_margin	15	Number of random bases between upstream adapter and template.
--total_length	500	Total read length in bases.
--number_of_sequences	50	Number of reads (or read pairs) to simulate.
--decay_model	"gaussian"	Quality decay model (gaussian, exponential, sigmoidal).
--decay_rate	0.0008	Steepness parameter for exponential ($1.25$ exponent) or sigmoidal drop.
--center	100	Peak center index (Gaussian) or drop position (Sigmoidal).
--std_dev	300	Standard deviation for Gaussian decay curve.
--min_val	40	Minimum quality score floor (clamped).
--max_val	73	Maximum quality score ceiling.
--score_type	"ascii"	Input format for min_val/max_val ("ascii" offset 33 or "phred" absolute).
--noise_level	0.5	Standard deviation of additive Gaussian noise on Phred scores.
--binned_quality	False	Snap quality scores to Illumina 4-state bins (Q2, Q12, Q23, Q37).
--homopolymer_penalty	False	Subtract $10$ Phred points within homopolymer runs $>4$ bases.

---

Testing

Run the test suite with pytest:

pytest tests/

If fastp is installed on your system PATH, integration tests will automatically execute fastp quality filtering verification.

---

License

This project is licensed under the MIT License. See LICENSE for details.

References

1. Cock PJ, Fields CJ, Goto N, Heuer ML, Rice PM. The Sanger FASTQ file format for sequences with quality scores, and the Solexa/Illumina FASTQ variants. Nucleic Acids Research. 2010;38(6):1767-1771. DOI: 10.1093/nar/gkp1137