Scaling NMR Cryoporometry Processing

The Hidden Difficulty of Experimental Data

Although NMR cryoporometry is a powerful technique for analysing pore structures in porous carbon systems, scaling data processing beyond a handful of samples introduces significant experimental and computational complexity. Raw instrument exports often contain inconsistent formatting, variable thermal ramp behaviour, sensor misalignment, unstable phase transition regions, and incomplete captures, all of which can introduce uncertainty into downstream interpretation and make reproducibility difficult when relying on fragmented scripts or spreadsheet based workflows.

Workflow Design

To address these challenges, I developed a modular workflow architecture for automating cryoporometry analysis from raw .txt exports through to processed pore distributions and comparative visualisation.

Technical Approach

A central orchestration layer coordinates parsing, processing, visualisation, and statistical export within one reproducible workflow.

workflow = NMRWorkflow(
    data_dir="./data",
    output_dir="./output",
    log_file="./output/nmr_processing.log"
)

results = workflow.run(
    generate_plots=True,
    generate_batch_summary=True
)

Experimental datasets are rarely perfectly structured, particularly across larger sample batches. Error handling was therefore integrated directly into the processing workflow to improve robustness when dealing with inconsistent formatting, unstable signal regions, or incomplete captures.

for sample_name, df in data_dict.items():
    try:
        processor = CryoporometryProcessor(df)
        processed = processor.process()

        results[sample_name] = processed

    except Exception as error:
        logger.error(f"{sample_name}: {error}")

The workflow automatically exports processed datasets, pore distribution outputs, and diagnostic visualisations for each sample, improving traceability and reducing repetitive manual preprocessing across experimental campaigns.

visualizer.plot_pore_distribution(
    pore_distribution,
    log_scale=True,
    output_path=f"{sample_name}_pore_distribution.png"
)

The full workflow implementation, processing modules, and visualisation pipeline are available on GitHub.

View the complete project code and documentation on my GitHub: sherlockyymh-prog/nmrdataprocessing

About me

Ongoing explorations

Contact

Location: Edinburgh