emit_sample_mzml/
emit_sample_mzml.rs

1//! Emit a small synthetic mzML document (plain and indexed) from the
2//! canonical writer, for schema-conformance checking in CI.
3//!
4//! Usage:
5//!   emit_sample_mzml [PLAIN_OUT] [INDEXED_OUT]
6//!
7//! Defaults to `sample_plain.mzML` and `sample_indexed.mzML` in the
8//! current directory. The output is validated against the vendored
9//! PSI-MS mzML XSD by the `validate-mzml` CI job.
10
11use std::fs::File;
12use std::io::BufWriter;
13
14use openmassspec_core::{
15    write_indexed_mzml, write_mzml, Activation, Analyzer, CvTerm, MobilityArrayKind, MsPower,
16    Polarity, PrecursorInfo, RunMetadata, ScanMode, SpectrumRecord, SpectrumSource,
17};
18
19struct SampleSource {
20    meta: RunMetadata,
21    spectra: Vec<SpectrumRecord>,
22    cursor: usize,
23}
24
25impl SampleSource {
26    fn new() -> Self {
27        let meta = RunMetadata {
28            source_file_name: "sample.raw".into(),
29            source_file_format: CvTerm::new("MS:1000563", "Thermo RAW format"),
30            native_id_format: CvTerm::new("MS:1000768", "Thermo nativeID format"),
31            instrument: CvTerm::new("MS:1001911", "Q Exactive"),
32            software_name: "openmassspec-core-sample".into(),
33            software_version: env!("CARGO_PKG_VERSION").into(),
34            start_timestamp: None,
35            mobility_array_kind: Some(MobilityArrayKind::InverseReducedVsPerCm2),
36        };
37        let ms1 = SpectrumRecord {
38            index: 0,
39            scan_number: 1,
40            native_id: "controllerType=0 controllerNumber=1 scan=1".into(),
41            ms_level: MsPower::Ms1.ms_level(),
42            polarity: Some(Polarity::Positive),
43            scan_mode: Some(ScanMode::Centroid),
44            analyzer: Some(Analyzer::FTMS),
45            filter: Some("FTMS + p ESI Full ms".into()),
46            retention_time_sec: 0.123 * 60.0,
47            total_ion_current: None,
48            base_peak_mz: None,
49            base_peak_intensity: None,
50            low_mz: None,
51            high_mz: None,
52            ion_injection_time_ms: Some(20.0),
53            inv_mobility: None,
54            precursor: None,
55            mz: vec![100.0, 200.0, 300.0],
56            intensity: vec![1.0, 5.0, 2.0],
57            inv_mobility_per_peak: None,
58        };
59        let ms2 = SpectrumRecord {
60            index: 1,
61            scan_number: 2,
62            native_id: "controllerType=0 controllerNumber=1 scan=2".into(),
63            ms_level: MsPower::Ms2.ms_level(),
64            polarity: Some(Polarity::Positive),
65            scan_mode: Some(ScanMode::Centroid),
66            analyzer: Some(Analyzer::FTMS),
67            filter: Some("FTMS + p ESI d Full ms2 200.00@hcd28.00".into()),
68            retention_time_sec: 0.5 * 60.0,
69            total_ion_current: Some(123.45),
70            base_peak_mz: Some(150.5),
71            base_peak_intensity: Some(99.0),
72            low_mz: Some(100.0),
73            high_mz: Some(180.0),
74            ion_injection_time_ms: Some(50.0),
75            inv_mobility: None,
76            precursor: Some(PrecursorInfo {
77                target_mz: Some(200.0),
78                selected_mz: Some(200.001),
79                isolation_width: Some(2.0),
80                charge: Some(2),
81                intensity: None,
82                collision_energy: Some(28.0),
83                ce_is_nce: true,
84                precursor_native_id: Some("controllerType=0 controllerNumber=1 scan=1".into()),
85                activation: Some(Activation::CID),
86                analyzer: Some(Analyzer::FTMS),
87            }),
88            mz: vec![150.5, 160.0],
89            intensity: vec![99.0, 50.0],
90            inv_mobility_per_peak: None,
91        };
92        // A frame-collapsed ion-mobility MS1 (the Bruker timsTOF shape):
93        // scalar 1/K0 plus a per-peak inverse-reduced-mobility array, which
94        // the writer emits as a third binary data array. Exercises the
95        // ion-mobility CV path so it is schema-checked without a vendor file.
96        let ms1_mobility = SpectrumRecord {
97            index: 2,
98            scan_number: 3,
99            native_id: "frame=1 scan=0".into(),
100            ms_level: MsPower::Ms1.ms_level(),
101            polarity: Some(Polarity::Positive),
102            scan_mode: Some(ScanMode::Centroid),
103            analyzer: Some(Analyzer::TOFMS),
104            filter: None,
105            retention_time_sec: 0.75 * 60.0,
106            total_ion_current: None,
107            base_peak_mz: None,
108            base_peak_intensity: None,
109            low_mz: None,
110            high_mz: None,
111            ion_injection_time_ms: None,
112            inv_mobility: Some(0.95),
113            precursor: None,
114            mz: vec![120.0, 240.0, 360.0],
115            intensity: vec![3.0, 7.0, 4.0],
116            inv_mobility_per_peak: Some(vec![0.92, 0.95, 0.98]),
117        };
118        Self {
119            meta,
120            spectra: vec![ms1, ms2, ms1_mobility],
121            cursor: 0,
122        }
123    }
124}
125
126impl SpectrumSource for SampleSource {
127    fn run_metadata(&self) -> RunMetadata {
128        self.meta.clone()
129    }
130
131    fn iter_spectra<'a>(&'a mut self) -> Box<dyn Iterator<Item = SpectrumRecord> + 'a> {
132        self.cursor = 0;
133        Box::new(std::iter::from_fn(move || {
134            let rec = self.spectra.get(self.cursor).cloned();
135            if rec.is_some() {
136                self.cursor += 1;
137            }
138            rec
139        }))
140    }
141
142    fn spectrum_count_hint(&self) -> Option<usize> {
143        Some(self.spectra.len())
144    }
145}
146
147fn main() -> std::io::Result<()> {
148    let mut args = std::env::args().skip(1);
149    let plain_path = args.next().unwrap_or_else(|| "sample_plain.mzML".into());
150    let indexed_path = args.next().unwrap_or_else(|| "sample_indexed.mzML".into());
151
152    let mut src = SampleSource::new();
153    let mut plain = BufWriter::new(File::create(&plain_path)?);
154    write_mzml(&mut src, &mut plain)
155        .map_err(|e| std::io::Error::other(format!("write_mzml: {e}")))?;
156
157    let mut indexed = BufWriter::new(File::create(&indexed_path)?);
158    write_indexed_mzml(&mut src, &mut indexed)
159        .map_err(|e| std::io::Error::other(format!("write_indexed_mzml: {e}")))?;
160
161    eprintln!("wrote {plain_path} and {indexed_path}");
162    Ok(())
163}
emit_sample_mzml/emit_sample_mzml.rs

emit_sample_mzml/
emit_sample_mzml.rs
