Add MS2 spectrum reading

src/lib.rs

@@ -2,6 +2,7 @@ mod file_types;
 mod parse_mzdata;
 mod parse_timsrust;
 mod precursor;
+mod ms2_spectrum;
 
 use std::collections::HashMap;
 
@@ -10,6 +11,7 @@ use pyo3::prelude::*;
 
 use file_types::{match_file_type, SpectrumFileType};
 use precursor::Precursor;
+use ms2_spectrum::MS2Spectrum;
 
 /// Get mapping of spectrum identifiers to precursor information.
 #[pyfunction]
@@ -31,11 +33,33 @@ pub fn get_precursor_info(spectrum_path: String) -> PyResult<HashMap<String, Pre
  }
 }
 
+/// Get MS2 spectra from a spectrum file.
+#[pyfunction]
+pub fn get_ms2_spectra(spectrum_path: String) -> PyResult<Vec<ms2_spectrum::MS2Spectrum>> {
+ let file_type = match_file_type(&spectrum_path);
+
+ let spectra = match file_type {
+ SpectrumFileType::MascotGenericFormat | SpectrumFileType::MzML => {
+ parse_mzdata::read_ms2_spectra(&spectrum_path, file_type)
+ }
+ SpectrumFileType::BrukerRaw => parse_timsrust::read_ms2_spectra(&spectrum_path),
+ // SpectrumFileType::ThermoRaw => parse_with_mzdata_thermo(&spectrum_path, file_type),
+ SpectrumFileType::Unknown => return Err(PyOSError::new_err("Unsupported file type")),
+ };
+
+ match spectra {
+ Ok(spectra) => Ok(spectra),
+ Err(e) => Err(PyOSError::new_err(e.to_string())),
+ }
+}
+
 
 /// A Python module implemented in Rust.
 #[pymodule]
 fn ms2rescore_rs(_py: Python, m: &PyModule) -> PyResult<()> {
  m.add_class::<Precursor>()?;
+ m.add_class::<MS2Spectrum>()?;
  m.add_function(wrap_pyfunction!(get_precursor_info, m)?)?;
+ m.add_function(wrap_pyfunction!(get_ms2_spectra, m)?)?;
  Ok(())
 }

src/ms2_spectrum.rs

@@ -0,0 +1,38 @@
+use pyo3::prelude::*;
+
+use crate::precursor::Precursor;
+
+#[pyclass(get_all, set_all)]
+#[derive(Debug, Clone)]
+pub struct MS2Spectrum {
+ pub identifier: String,
+ pub mz: Vec<f32>,
+ pub intensity: Vec<f32>,
+ pub precursor: Option<Precursor>,
+}
+
+impl MS2Spectrum {
+ pub fn new(
+ identifier: String,
+ mz: Vec<f32>,
+ intensity: Vec<f32>,
+ precursor: Option<Precursor>,
+ ) -> Self {
+ MS2Spectrum {
+ identifier,
+ mz,
+ intensity,
+ precursor,
+ }
+ }
+}
+
+#[pymethods]
+impl MS2Spectrum {
+ fn __repr__(&self) -> String {
+ format!(
+ "MS2Spectrum(identifier='{}', mz=[..], intensity=[..], precursor={:?})",
+ self.identifier, self.precursor
+ )
+ }
+}

src/parse_mzdata.rs

@@ -4,10 +4,11 @@ use std::fs::File;
 use mzdata::io::{MGFReader, MzMLReader};
 
 use crate::file_types::SpectrumFileType;
+use crate::ms2_spectrum::MS2Spectrum;
 use crate::precursor::Precursor;
 
-impl From<mzdata::spectrum::MultiLayerSpectrum> for Precursor {
- fn from(spectrum: mzdata::spectrum::MultiLayerSpectrum) -> Self {
+impl From<&mzdata::spectrum::MultiLayerSpectrum> for Precursor {
+ fn from(spectrum: &mzdata::spectrum::MultiLayerSpectrum) -> Self {
  let precursor = &spectrum.description.precursor;
  match precursor {
  Some(precursor) => Precursor {
@@ -18,18 +19,33 @@ impl From<mzdata::spectrum::MultiLayerSpectrum> for Precursor {
  .first_scan()
  .map(|s| s.start_time)
  .unwrap_or(0.0),
- im: get_im_from_spectrum_description(&spectrum)
- .or(get_im_from_selected_ion(&spectrum))
- .or(get_im_from_first_scan(&spectrum))
+ im: get_im_from_spectrum_description(spectrum)
+ .or(get_im_from_selected_ion(spectrum))
+ .or(get_im_from_first_scan(spectrum))
  .unwrap_or(0.0),
- charge: get_charge_from_spectrum(&spectrum).unwrap_or(0),
+ charge: get_charge_from_spectrum(spectrum).unwrap_or(0),
  intensity: precursor.ions[0].intensity as f64,
  },
  None => Precursor::default(),
  }
  }
 }
 
+impl From<mzdata::spectrum::MultiLayerSpectrum> for MS2Spectrum {
+ fn from(spectrum: mzdata::spectrum::MultiLayerSpectrum) -> Self {
+ let identifier: String = spectrum.description.id.to_string();
+ let precursor = Precursor::from(&spectrum);
+ let mzdata_centroid_spectrum = spectrum.into_centroid().unwrap();
+ let (mz, intensity): (Vec<f32>, Vec<f32>) = mzdata_centroid_spectrum
+ .peaks
+ .iter()
+ .map(|peak| (peak.mz as f32, peak.intensity))
+ .unzip();
+
+ MS2Spectrum::new(identifier, mz, intensity, Some(precursor))
+ }
+}
+
 /// Parse precursor info from spectrum files with mzdata
 pub fn parse_precursor_info(
  spectrum_path: &str,
@@ -40,7 +56,7 @@ pub fn parse_precursor_info(
  SpectrumFileType::MascotGenericFormat => Ok(MGFReader::new(file)
  .filter_map(|spectrum| {
  spectrum.description.precursor.as_ref()?;
- Some((spectrum.description.id.clone(), Precursor::from(spectrum)))
+ Some((spectrum.description.id.clone(), Precursor::from(&spectrum)))
  })
  .collect::<HashMap<String, Precursor>>()),
 
@@ -50,7 +66,7 @@ pub fn parse_precursor_info(
  return None;
  }
  spectrum.description.precursor.as_ref()?;
- Some((spectrum.description.id.clone(), Precursor::from(spectrum)))
+ Some((spectrum.description.id.clone(), Precursor::from(&spectrum)))
  })
  .collect::<HashMap<String, Precursor>>()),
 
@@ -61,6 +77,29 @@ pub fn parse_precursor_info(
  }
 }
 
+/// Read MS2 spectra from spectrum files with mzdata
+pub fn read_ms2_spectra(
+ spectrum_path: &str,
+ file_type: SpectrumFileType,
+) -> Result<Vec<MS2Spectrum>, std::io::Error> {
+ let file = File::open(spectrum_path)?;
+ match file_type {
+ SpectrumFileType::MascotGenericFormat => Ok(MGFReader::new(file)
+ .map(MS2Spectrum::from)
+ .collect::<Vec<MS2Spectrum>>()),
+
+ SpectrumFileType::MzML => Ok(MzMLReader::new(file)
+ .filter(|spectrum| spectrum.description.ms_level == 2)
+ .map(MS2Spectrum::from)
+ .collect::<Vec<MS2Spectrum>>()),
+
+ _ => Err(std::io::Error::new(
+ std::io::ErrorKind::InvalidInput,
+ "Unsupported file type for mzdata",
+ )),
+ }
+}
+
 // pub fn parse_precursor_info_thermo(
 // spectrum_path: &str,
 // file_type: SpectrumFileType,

src/parse_timsrust.rs

@@ -1,5 +1,6 @@
 use std::collections::HashMap;
 
+use crate::ms2_spectrum::MS2Spectrum;
 use crate::precursor::Precursor;
 
 impl From<timsrust::Precursor> for Precursor {
@@ -14,6 +15,24 @@ impl From<timsrust::Precursor> for Precursor {
  }
 }
 
+impl From<timsrust::Spectrum> for MS2Spectrum {
+ fn from(spectrum: timsrust::Spectrum) -> Self {
+ MS2Spectrum::new(
+ spectrum.index.to_string(),
+ spectrum.mz_values.iter().map(|mz| *mz as f32).collect(),
+ spectrum
+ .intensities
+ .iter()
+ .map(|intensity| *intensity as f32)
+ .collect(),
+ match spectrum.precursor {
+ timsrust::QuadrupoleEvent::Precursor(precursor) => Some(Precursor::from(precursor)),
+ _ => None,
+ },
+ )
+ }
+}
+
 /// Parse precursor info from spectrum files with timsrust
 pub fn parse_precursor_info(
  spectrum_path: &str,
@@ -38,3 +57,15 @@ pub fn parse_precursor_info(
  })
  .collect::<HashMap<String, Precursor>>())
 }
+
+/// Read MS2 spectra from spectrum files with timsrust
+pub fn read_ms2_spectra(spectrum_path: &str) -> Result<Vec<MS2Spectrum>, std::io::Error> {
+ let reader = timsrust::FileReader::new(spectrum_path)
+ .map_err(|e| std::io::Error::new(std::io::ErrorKind::Other, e.to_string()))?;
+
+ Ok(reader
+ .read_all_spectra()
+ .into_iter()
+ .map(MS2Spectrum::from)
+ .collect())
+}

src/precursor.rs

@@ -2,7 +2,7 @@ use pyo3::prelude::*;
 
 /// Precursor information.
 #[pyclass(get_all, set_all)]
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 pub struct Precursor {
  pub mz: f64,
  pub rt: f64,