Merge pull request #11 from MannLabs/develop

Develop

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "timsrust"
-version = "0.2.1"
+version = "0.2.2"
 edition = "2021"
 description = "A crate to read Bruker timsTOF data"
 license = "Apache-2.0"

README.md

@@ -1,3 +1,7 @@
+
+![Crates.io](https://img.shields.io/crates/v/timsrust?link=https%3A%2F%2Fcrates.io%2Fcrates%2Ftimsrust)
+![docs.rs](https://img.shields.io/docsrs/timsrust?link=https%3A%2F%2Fdocs.rs%2Ftimsrust%2F0.2.1%2Ftimsrust%2F)
+
 # TimsRust
 
 A crate to read Bruker TimsTof data.
@@ -24,12 +28,17 @@ Two primary data types are exposed through TimsRust:
 ### File formats
 
 Two file formats are supported:
-* Bruker .d folder containing:
+* TDF - Bruker .d folder containing:
     * analysis.tdf
     * analysis.tdf_bin
-* Bruker .ms2 folder containing:
-    * converter.ms2.bin
-    * converter.MS2Spectra.ms2.parquet
+
+* miniTDF - ProteoScape optimized Bruker file-format. Similar to TDF, miniTDF consists of multiple files: a binary '.bin'
+ and an index '.parquet' file. The file-names are made up to the following convention: `<producing-engine-name>.<domain-name>.<extension>`.
+ e.g. for MS2 spectrum information: `<producing-engine-name>.ms2spectrum.<extension>`. Therefore the following files are expected
+ in the provided ms2 folder:
+    * *.ms2spectrum.bin
+    * *.ms2spectrum.parquet
+
 
 ## Python bindings
 

src/converters.rs

@@ -1,9 +1,12 @@
 use linreg::linear_regression;
 
+/// Converting from an index domain (e.g. Time of Flight) to a continuous domain (m/z).
 pub trait ConvertableIndex {
+    /// Convert any index (even fractional) to a continuous value.
     fn convert<T: Into<f64> + Copy>(&self, index: T) -> f64;
 }
 
+/// A converter from TOF -> m/z.
 #[derive(Debug, Copy, Clone)]
 pub struct Tof2MzConverter {
     tof_intercept: f64,
@@ -40,6 +43,7 @@ impl ConvertableIndex for Tof2MzConverter {
     }
 }
 
+/// A converter from Scan -> ion mobility.
 #[derive(Debug, Copy, Clone)]
 pub struct Scan2ImConverter {
     scan_intercept: f64,
@@ -64,6 +68,7 @@ impl ConvertableIndex for Scan2ImConverter {
     }
 }
 
+/// A converter from Frame -> retention time.
 #[derive(Debug, Clone)]
 pub struct Frame2RtConverter {
     rt_values: Vec<f64>,

src/file_readers.rs

@@ -1,4 +1,7 @@
-use crate::Error;
+use crate::{
+    converters::{Frame2RtConverter, Scan2ImConverter, Tof2MzConverter},
+    Error,
+};
 
 mod common;
 mod file_formats;
@@ -15,6 +18,8 @@ use {
 
 pub use file_formats::FileFormatError;
 
+use self::frame_readers::tdf_reader::TDFReader;
+
 /// A reader to read [frames](crate::Frame) and [spectra](crate::Spectrum).
 pub struct FileReader {
     format: FileFormat,
@@ -37,10 +42,14 @@ impl FileReader {
         self.format.read_all_frames()
     }
 
+    /// NOTE: The returned vec contains all frames to not disrupt indexing.
+    /// MS2 frames are set to unknown and not read.
     pub fn read_all_ms1_frames(&self) -> Vec<Frame> {
         self.format.read_all_ms1_frames()
     }
 
+    /// NOTE: The returned vec contains all frames to not disrupt indexing.
+    /// MS1 frames are set to unknown and not read.
     pub fn read_all_ms2_frames(&self) -> Vec<Frame> {
         self.format.read_all_ms2_frames()
     }
@@ -56,4 +65,40 @@ impl FileReader {
     pub fn read_all_spectra(&self) -> Vec<Spectrum> {
         self.format.read_all_spectra()
     }
+
+    pub fn get_frame_converter(&self) -> Result<Frame2RtConverter, Error> {
+        match &self.format {
+            FileFormat::DFolder(path) => Ok(TDFReader::new(
+                &path.to_str().unwrap_or_default().to_string(),
+            )
+            .rt_converter),
+            _ => Err(Error::FileFormatError(
+                FileFormatError::MetadataFilesAreMissing,
+            )),
+        }
+    }
+
+    pub fn get_scan_converter(&self) -> Result<Scan2ImConverter, Error> {
+        match &self.format {
+            FileFormat::DFolder(path) => Ok(TDFReader::new(
+                &path.to_str().unwrap_or_default().to_string(),
+            )
+            .im_converter),
+            _ => Err(Error::FileFormatError(
+                FileFormatError::MetadataFilesAreMissing,
+            )),
+        }
+    }
+
+    pub fn get_tof_converter(&self) -> Result<Tof2MzConverter, Error> {
+        match &self.format {
+            FileFormat::DFolder(path) => Ok(TDFReader::new(
+                &path.to_str().unwrap_or_default().to_string(),
+            )
+            .mz_converter),
+            _ => Err(Error::FileFormatError(
+                FileFormatError::MetadataFilesAreMissing,
+            )),
+        }
+    }
 }

src/file_readers/file_formats.rs

@@ -20,15 +20,7 @@ impl FileFormat {
             .unwrap_or_default();
         let format = match extension {
             "d" => Self::DFolder(path),
-            "ms2" => Self::MS2Folder(path),
-            _ => {
-                if let Some(path) = path.parent() {
-                    // Only recurse if there is a valid parent section,
-                    // otherwise we'll get a stack overflow
-                    return Self::parse(path);
-                }
-                return Err(FileFormatError::NoParentWithBrukerExtension);
-            },
+            _ => Self::MS2Folder(path),
         };
         format.is_valid()?;
         Ok(format)

src/file_readers/spectrum_readers/mini_tdf_reader.rs

@@ -1,3 +1,5 @@
+use crate::file_readers::FileFormatError;
+use std::fs;
 use {
     crate::{
         file_readers::{
@@ -7,7 +9,7 @@ use {
             },
             ReadableSpectra,
         },
-        precursors::PrecursorType,
+        precursors::QuadrupoleEvent,
         Precursor, Spectrum,
     },
     rayon::prelude::*,
@@ -23,6 +25,41 @@ pub struct MiniTDFReader {
     frame_reader: BinFileReader,
 }
 
+fn find_ms2spectrum_file(
+    ms2_dir_path: &str,
+    extension: String,
+) -> Result<String, FileFormatError> {
+    let files = fs::read_dir(ms2_dir_path).unwrap();
+    for file in files {
+        let filename = file
+            .unwrap()
+            .path()
+            .file_name()
+            .unwrap()
+            .to_str()
+            .unwrap()
+            .to_owned();
+        if filename
+            .ends_with(std::format!("ms2spectrum.{}", extension).as_str())
+        {
+            return Ok(filename);
+        }
+    }
+    let err = match extension.as_str() {
+        "parquet" => FileFormatError::MetadataFilesAreMissing,
+        "bin" => FileFormatError::BinaryFilesAreMissing,
+        _ => FileFormatError::BinaryFilesAreMissing,
+    };
+    println!(
+        "{}",
+        format!(
+            "No '*.ms2spectrum.{}' file found in '{}'",
+            extension, ms2_dir_path
+        )
+    );
+    return Err(err);
+}
+
 impl MiniTDFReader {
     pub fn new(path_name: String) -> Self {
         let mut reader: MiniTDFReader = Self::default();
@@ -35,7 +72,10 @@ impl MiniTDFReader {
 
     fn read_parquet_file_name(&mut self) {
         let mut path: PathBuf = PathBuf::from(&self.path_name);
-        path.push("converter.MS2Spectra.ms2.parquet");
+        let ms2_parquet_file =
+            find_ms2spectrum_file(&self.path_name, "parquet".to_owned())
+                .unwrap();
+        path.push(ms2_parquet_file);
         self.parquet_file_name = path.to_string_lossy().into_owned();
     }
 
@@ -45,7 +85,9 @@ impl MiniTDFReader {
     }
     fn set_spectrum_reader(&mut self) {
         let mut path: PathBuf = PathBuf::from(&self.path_name);
-        path.push("converter.ms2.bin");
+        let ms2_bin_file =
+            find_ms2spectrum_file(&self.path_name, "bin".to_owned()).unwrap();
+        path.push(ms2_bin_file);
         let file_name: String = path.to_string_lossy().into_owned();
         self.frame_reader =
             BinFileReader::new(String::from(&file_name), self.offsets.clone());
@@ -56,7 +98,7 @@ impl ReadableSpectra for MiniTDFReader {
     fn read_single_spectrum(&self, index: usize) -> Spectrum {
         let mut spectrum: Spectrum =
             Spectrum::read_from_file(&self.frame_reader, index);
-        spectrum.precursor = PrecursorType::Precursor(self.precursors[index]);
+        spectrum.precursor = QuadrupoleEvent::Precursor(self.precursors[index]);
         spectrum
     }
 

src/lib.rs

@@ -11,12 +11,15 @@
 //!
 //! Two file formats are supported:
 //!
-//! * Bruker .ms2 folder containing:
-//!     * converter.ms2.bin
-//!     * converter.MS2Spectra.ms2.parquet
 //! * Bruker .d folder containing:
 //!     * analysis.tdf
 //!     * analysis.tdf_bin
+//! * miniTDF - ProteoScape optimized Bruker file-format. Similar to TDF, miniTDF consists of multiple files: a binary '.bin'
+//!  and an index '.parquet' file. The file-names are made up to the following convention: `<producing-engine-name>.<domain-name>.<extension>`.
+//!  e.g. for MS2 spectrum information: `<producing-engine-name>.ms2spectrum.<extension>`. Therefore the following files are expected
+//!  in the provided ms2 folder:
+//!     * *.ms2spectrum.bin
+//!     * *.ms2spectrum.parquet
 
 mod acquisition;
 mod calibration;
@@ -30,9 +33,12 @@ mod vec_utils;
 
 pub use crate::{
     acquisition::AcquisitionType,
+    converters::{
+        ConvertableIndex, Frame2RtConverter, Scan2ImConverter, Tof2MzConverter,
+    },
     errors::*,
     file_readers::FileReader,
     frames::{Frame, FrameType},
-    precursors::{Precursor, PrecursorType},
+    precursors::{Precursor, QuadrupoleEvent},
     spectra::Spectrum,
 };

src/precursors.rs

@@ -12,23 +12,23 @@ pub struct Precursor {
 
 /// A type of quadrupole selection.
 #[derive(Debug, Clone, Copy, PartialEq)]
-pub enum PrecursorType {
+pub enum QuadrupoleEvent {
     Precursor(Precursor),
     // Window(Window),
     // PrecursorList(Vec<Precursor>),
     None,
 }
 
-impl Default for PrecursorType {
+impl Default for QuadrupoleEvent {
     fn default() -> Self {
         Self::None
     }
 }
 
-impl PrecursorType {
+impl QuadrupoleEvent {
     pub fn unwrap_as_precursor(&self) -> Precursor {
         match self {
-            PrecursorType::Precursor(precursor) => *precursor,
+            QuadrupoleEvent::Precursor(precursor) => *precursor,
             _ => {
                 panic!("Not a precursor");
             },

src/spectra.rs

@@ -1,6 +1,6 @@
 use crate::{
     converters::{ConvertableIndex, Tof2MzConverter},
-    precursors::PrecursorType,
+    precursors::QuadrupoleEvent,
     vec_utils::{filter_with_mask, find_sparse_local_maxima_mask},
     Precursor,
 };
@@ -78,7 +78,7 @@ impl RawSpectrumProcessor {
                 .iter()
                 .map(|x| *x as f64)
                 .collect(),
-            precursor: PrecursorType::Precursor(precursor),
+            precursor: QuadrupoleEvent::Precursor(precursor),
             index: index,
         };
         spectrum
@@ -112,6 +112,6 @@ pub struct RawSpectrum {
 pub struct Spectrum {
     pub mz_values: Vec<f64>,
     pub intensities: Vec<f64>,
-    pub precursor: PrecursorType,
+    pub precursor: QuadrupoleEvent,
     pub index: usize,
 }

tests/simulation_requirements.txt

@@ -0,0 +1,4 @@
+pandas
+numpy
+pyzstd
+ipykernel