Skip to content
BranchesTags

Repository files navigation

Regolyx

DOI License Python Status Platform

Codon Optimization and Structural Modeling of (Per)chlorate-Reducing Enzymes from Three Bacterial Donors for Bacillus subtilis-Based Martian Regolith Bioremediation

Overview

Regolyx is a reproducible bioinformatics workflow for the computational engineering of perchlorate-reducing enzymes intended for expression in Bacillus subtilis as a potential chassis organism for Martian regolith bioremediation.

Perchlorates are widespread contaminants in Martian regolith and represent a major challenge for future human habitation and in-situ resource utilization (ISRU). This project investigates whether key enzymes from naturally occurring perchlorate-reducing bacteria can be optimized for heterologous expression in B. subtilis while retaining favorable structural characteristics.

The workflow integrates sequence verification, codon optimization, secondary-structure analysis, and homology modeling into a single reproducible pipeline.

See CHANGELOG.md for a full record of corrections made since the initial release, including a gene/accession verification failure that was identified and resolved (summary below).

Key Results

A total of nine perchlorate- and chlorate-reduction pathway enzymes from three bacterial donors were optimized for expression in Bacillus subtilis and structurally assessed using SWISS-MODEL.

Codon Optimization Summary

Gene Protein Length (aa) DNA Length (bp) GC (%) CAI
Cld_Azos 282 846 37.59 1.000
Cld_Dech 282 846 37.71 1.000
Cld_Ideo 285 855 40.47 1.000
ClrA_Ideo 914 2742 40.15 1.000
ClrB_Ideo 328 984 42.58 1.000
PcrA_Azos 927 2781 40.27 1.000
PcrA_Dech 927 2781 40.52 1.000
PcrB_Azos 333 999 43.64 1.000
PcrB_Dech 333 999 43.54 1.000

CAI is mathematically fixed at 1.000 for every sequence by construction: the optimization method always substitutes the single highest-frequency synonymous codon, so CAI against the same reference table cannot vary. This is reported as a stated property of the optimization method, not as a measure of result quality. D. aromatica and A. suillum PcrA/PcrB pairs share identical lengths but are non-identical sequences (88.0% and 94.6% pairwise identity respectively), consistent with orthologous genes between related species.

RNA Folding Analysis

Gene MFE (kcal/mol)
Cld_Azos 0.00
Cld_Dech -4.00
Cld_Ideo -2.40
ClrA_Ideo -0.20
ClrB_Ideo -2.80
PcrA_Azos -1.50
PcrA_Dech -1.50
PcrB_Azos -1.20
PcrB_Dech -1.20

Values are minimum free energy for the first 30nt of each optimized coding sequence (translation-initiation region), predicted with ViennaRNA at 37°C. An MFE of 0.00 (Cld_Azos) reflects a genuinely unstructured window in that prediction, not a missing value.

Structural Modeling Summary

Gene GMQE QMEAN Status
Cld_Azos 0.82 0.91 COMPLETED
Cld_Dech 0.83 0.91 COMPLETED
Cld_Ideo 0.75 0.85 COMPLETED
ClrA_Ideo 0.70 0.74 COMPLETED
ClrB_Ideo 0.84 0.78 COMPLETED
PcrA_Azos 0.93 0.95 COMPLETED
PcrA_Dech 0.91 0.93 COMPLETED
PcrB_Azos 0.99 0.95 COMPLETED
PcrB_Dech 0.98 0.94 COMPLETED

GMQE and QMEAN (average local score) are retrieved directly from the SWISS-MODEL REST API's structured JSON response. Template identity and coverage are not returned by this API endpoint and are not reported here; see Known Limitations.

Verification Summary

All nine constructs:

  • Passed accession verification against primary structural (PDB) and curated (UniProt/Swiss-Prot) database records, independently re-verified by direct inspection
  • Retained full-length coding sequences
  • Produced zero unmapped residues during codon optimization
  • Achieved CAI = 1.000 (see note above on why this is expected, not a quality signal)
  • Successfully completed SWISS-MODEL processing via the REST API

Note on accession provenance: an earlier accession set used in this pipeline was found, on independent re-verification, to be incorrect for five of the nine genes (each resolving to a real but unrelated protein from an unrelated organism). All nine accessions were re-sourced and re-verified before this release; see docs/verification_log.md for the corrected accession table and CHANGELOG.md for the full incident record.

Highlights

  • Verification of experimentally characterized perchlorate-reduction genes, independently re-confirmed against primary database records
  • Codon optimization for Bacillus subtilis
  • Secondary-structure analysis of optimized transcripts
  • Automated SWISS-MODEL structural modeling via direct REST API integration (no manual upload or HTML scraping)
  • Model quality assessment and reporting from structured API data
  • Fully reproducible computational workflow

Research Objectives

  • Verify target enzyme sequences from perchlorate-reducing bacteria
  • Optimize coding sequences for expression in Bacillus subtilis
  • Evaluate codon adaptation and sequence composition
  • Assess RNA secondary-structure characteristics
  • Generate homology models using SWISS-MODEL
  • Compare model quality metrics across donor organisms
  • Produce publication-ready summary reports

Target Enzymes

Donor Organism Gene Function Accession Notes
Azospira suillum PS PcrA Catalytic perchlorate reductase subunit UniProt G8QM55 Strain-specific (PDB 5E7O, locus Dsui_0149)
Azospira suillum PS PcrB Electron-transfer subunit UniProt G8QM54 Strain-specific (PDB 5E7O, locus Dsui_0148)
Azospira suillum PS Cld Chlorite dismutase UniProt E2DI02 From strain GR-1, not PS — no PS-specific record available
Dechloromonas aromatica RCB PcrA Catalytic perchlorate reductase subunit UniProt Q47CW6 Strain-specific (Daro_2584)
Dechloromonas aromatica RCB PcrB Electron-transfer subunit UniProt Q47CW7 Strain-specific (Daro_2583)
Dechloromonas aromatica RCB Cld Chlorite dismutase UniProt Q47CX0 Strain-specific (Daro_2580)
Ideonella dechloratans ClrA Chlorate reductase catalytic subunit UniProt P60068 Species-specific
Ideonella dechloratans ClrB Electron-transfer subunit UniProt P60069 Species-specific
Ideonella dechloratans Cld Chlorite dismutase UniProt Q9F437 Species-specific

Full provenance and verification method for each accession are documented in docs/verification_log.md.

Workflow

Sequence Verification
        |
        v
Codon Usage Analysis
        |
        v
Codon Optimization
        |
        v
Secondary Structure Prediction
        |
        v
SWISS-MODEL Homology Modeling (REST API)
        |
        v
Model Quality Assessment
        |
        v
Automated Reporting

Repository Structure

regolyx/
├── CITATION.cff
├── CHANGELOG.md
├── LICENSE
├── README.md
├── environment.yml
├── requirements.txt
├── data/
│   ├── bsubtilis_codon_usage.csv
│   └── sequences/
├── docs/
│   └── verification_log.md
├── scripts/
│   ├── step0_verify_accessions.py
│   ├── step1_codon_usage.py
│   ├── step2_codon_optimize.py
│   ├── step3_secondary_structure.py
│   ├── step4_swissmodel_api.py
│   └── step5_generate_report.py
└── results/
    ├── codon_optimization/
    ├── secondary_structure/
    └── structural_modeling/

Installation

Conda

conda env create -f environment.yml
conda activate regolyx

Pip

pip install -r requirements.txt

Running the Pipeline

python scripts/step0_verify_accessions.py
python scripts/step1_codon_usage.py
python scripts/step2_codon_optimize.py
python scripts/step3_secondary_structure.py
python scripts/step4_swissmodel_api.py   # requires SWISSMODEL_TOKEN env var; see script docstring
python scripts/step5_generate_report.py

Outputs

Codon Optimization

Location:

results/codon_optimization/

Includes:

  • Optimized coding sequences
  • Codon usage statistics
  • Summary tables

Secondary Structure

Location:

results/secondary_structure/

Includes:

  • RNA folding predictions
  • Structure summaries

Structural Modeling

Location:

results/structural_modeling/

Includes:

  • SWISS-MODEL predictions (retrieved via REST API as structured JSON, swissmodel_api_results.json)
  • Model quality metrics (model_quality_summary.csv)
  • Structural summaries

Known Limitations

  • Template identity and coverage are not reported. The SWISS-MODEL REST API endpoint used by this pipeline returns GMQE and QMEAN (average local score) but does not return template sequence identity, coverage, or template ID in its response. These fields are present as empty columns in model_quality_summary.csv rather than omitted, so the gap is visible. A future revision may query a separate SWISS-MODEL endpoint if this data is required for downstream analysis.
  • Native-sequence CAI/GC baseline comparison is pending. The verified gene targets are protein sequences; no native nucleotide CDS is available for direct before/after comparison of codon optimization. Reverse-translation was deliberately not used as a substitute, since it would not reflect the donor organism's actual codon usage.
  • Cld for Azospira suillum PS is a cross-strain substitution. No strain-PS-specific chlorite dismutase accession was found in UniProt or NCBI. The sequence used (UniProt E2DI02) is from Azospira oryzae strain GR-1, the strain in which Cld was originally biochemically characterized for this genus. Any manuscript text or figure using this sequence should cite strain GR-1 explicitly, not strain PS.

Scientific Context

Keywords:

  • Perchlorate Reduction
  • Chlorite Dismutase
  • Perchlorate Reductase
  • Synthetic Biology
  • Bacillus subtilis
  • Codon Optimization
  • Structural Bioinformatics
  • SWISS-MODEL
  • Astrobiotechnology
  • Martian Regolith
  • ISRU
  • Bioremediation

Reproducibility

Environment:

  • Python 3.13
  • Conda environment specification included
  • Requirements file included

Pipeline stages:

  1. Sequence verification
  2. Codon usage analysis
  3. Codon optimization
  4. RNA secondary structure prediction
  5. SWISS-MODEL structural modeling (REST API)
  6. Automated reporting

All analyses can be reproduced from raw sequence files using the scripts contained in the scripts/ directory. scripts/step4_swissmodel_api.py requires a SWISS-MODEL account API token, set via the SWISSMODEL_TOKEN environment variable — see the script's docstring for how to obtain one. No password or token should ever be passed as a command-line argument or committed to this repository.

Data Availability

All input sequences, optimized constructs, structural models, intermediate files, and analysis outputs are available in this repository.

Archived release:

https://zenodo.org/records/20838989

DOI:

10.5281/zenodo.20838989

Note: the archived Zenodo release should be checked against the version of data/sequences/ it captured. If the archived snapshot predates the accession corrections described in CHANGELOG.md, a new Zenodo version should be minted against the corrected data before this DOI is cited as authoritative.

Citation

If you use Regolyx in academic work, please cite the metadata contained in:

CITATION.cff

CITATION.cff should be updated to reflect the current version (see CHANGELOG.md) before this repository is cited externally.

License

Released under the MIT License.

See LICENSE for details.

About

Codon optimization and structural modeling of perchlorate-reducing enzymes for Martian regolith bioremediation.

Topics

astrobiologybacillus-subtilisbioinformaticsbioremediationchloratecodon-optimizationmartian-regolithperchlorateprotein-modelingswiss-modelsynthetic-biologyviennarna

Resources

Readme
MIT license
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

Packages

Contributors

Languages