Quickstart: Rust

Add the umbrella crate to your Cargo.toml:

[dependencies]
openmassspec-io   = { path = "../OpenMassSpec/crates/openmassspec-io", features = ["all"] }
openmassspec-core = { path = "../OpenMassSpecCore" }

The all feature pulls in every vendor. To trim binary size you can opt in to one vendor at a time: features = ["thermo"], features = ["bruker"], features = ["waters"], features = ["agilent"], features = ["sciex"], features = ["shimadzu"].

Convert a file to mzML

use std::path::Path;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let input  = Path::new("sample.raw");
    let output = Path::new("sample.mzML");

    let detected = openmassspec_io::detect_format(input)
        .ok_or("not a recognized vendor format")?;
    openmassspec_io::convert_to_mzml(detected, output, /* indexed = */ true)?;
    Ok(())
}

Centroid on the way out

collect, convert_to_mzml, and convert_to_mzml_writer each have a _centroided sibling that peak-picks every profile-mode spectrum first (spectra already tagged centroid pass through unchanged) via openmassspec_core::Centroided. They take an extra min_intensity: Option<f32> noise floor. Plain collect / convert_to_mzml never centroid - this is always opt-in:

use std::path::Path;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let detected = openmassspec_io::detect_format(Path::new("sample.raw"))
        .ok_or("not a recognized vendor format")?;
    openmassspec_io::convert_to_mzml_centroided(
        detected,
        Path::new("sample.mzML"),
        /* indexed = */ true,
        /* min_intensity = */ Some(500.0),
    )?;
    Ok(())
}

Iterate spectra without writing mzML

use openmassspec_core::SpectrumSource;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let detected = openmassspec_io::detect_format(std::path::Path::new("sample.raw"))
        .ok_or("not a vendor format")?;
    let (records, meta) = openmassspec_io::collect(detected)?;
    println!("{} spectra from {}", records.len(), meta.instrument.name);
    for s in records.iter().take(5) {
        println!(
            "idx={} ms={} rt={:.2}s peaks={}",
            s.index, s.ms_level, s.retention_time_sec, s.mz.len()
        );
    }
    Ok(())
}

For long runs you usually want the streaming variant. Open the vendor source directly and drive iter_spectra yourself - this is what convert_to_mzml does internally:

use openmassspec_core::SpectrumSource;

let mut src = opentimstdf::mzml::TdfSource::open("sample.d")?;
for s in src.iter_spectra() {
    // process one spectrum at a time
}
# Ok::<(), Box<dyn std::error::Error>>(())

Validate

use openmassspec_core::conformance::assert_iter_invariants;

let detected = openmassspec_io::detect_format(std::path::Path::new("sample.raw")).unwrap();
let (records, _) = openmassspec_io::collect(detected)?;
let n = assert_iter_invariants(records.iter().cloned())?;
println!("conformance ok: {n} spectra");
# Ok::<(), Box<dyn std::error::Error>>(())

Arrow

Enable the arrow feature on openmassspec-core and build a record batch directly:

use openmassspec_core::arrow::SpectrumBatchBuilder;
use openmassspec_core::SpectrumSource;

let mut src = opentimstdf::mzml::TdfSource::open("sample.d")?;
let mut b = SpectrumBatchBuilder::new(None);
for s in src.iter_spectra() {
    b.push(&s);
}
let batch = b.finish()?;
println!("{} rows x {} cols", batch.num_rows(), batch.num_columns());
# Ok::<(), Box<dyn std::error::Error>>(())