Match those fragments!

Handle mass spectrometry data in Rust. This crate is set up to handle very complex peptides with loads of ambiguity and complexity. It pivots around the CompoundPeptidoform, Peptidoform and LinearPeptide which encode the ProForma specification. Additionally this crate enables the reading of mgf, doing spectrum annotation (BU/MD/TD), finding isobaric sequences, doing alignments of peptides , accessing the IMGT germline database, and reading identified peptide files.

Library features

• Read ProForma sequences (complete specification supported: 'level 2-ProForma + top-down compliant + cross-linking compliant + glycans compliant + mass spectrum compliant')
• Generate theoretical fragments with control over the fragmentation model from any ProForma peptidoform/proteoform
  • Generate theoretical fragments for chimeric spectra
  • Generate theoretical fragments for cross-links (also disulfides)
  • Generate theoretical fragments for modifications of unknown position
  • Generate peptide backbone (a, b, c, x, y, and z) and satellite ion fragments (w, d, and v)
  • Generate glycan fragments (B, Y, and internal fragments)
• Integrated with mzdata for reading raw data files
• Match spectra to the generated fragments
• Align peptides based on mass
• Fast access to the IMGT database of antibody germlines
• Reading of multiple identified peptide file formats (Fasta, MaxQuant, MSFragger, Novor, OPair, Peaks, and Sage)
• Exhaustively fuzz tested for reliability (using cargo-afl)
• Extensive use of uom for compile time unit checking

Example usage

# fn main() -> Result<(), rustyms::error::CustomError> {
# let raw_file_path = "data/annotated_example.mgf";
use rustyms::{*, system::{usize::Charge, e}};
// Open example raw data (this is the built in mgf reader, look into mzdata for more advanced raw file readers)
let spectrum = rawfile::mgf::open(raw_file_path)?;
// Parse the given ProForma definition
let peptide = CompoundPeptidoform::pro_forma("[Gln->pyro-Glu]-QVQEVSERTHGGNFD", None)?;
// Generate theoretical fragments for this peptide given EThcD fragmentation
let model = Model::ethcd();
let fragments = peptide.generate_theoretical_fragments(Charge::new::<e>(2), &model);
// Annotate the raw data with the theoretical fragments
let annotated = spectrum[0].annotate(peptide, &fragments, &model, MassMode::Monoisotopic);
// Calculate a peak false discovery rate for this annotation 
let (fdr, _) = annotated.fdr(&fragments, &model, MassMode::Monoisotopic);
// This is the incorrect sequence for this spectrum so the peak FDR will indicate this
# dbg!(&fdr, fdr.peaks_sigma(), fdr.peaks_fdr(), fdr.peaks_score());
assert!(fdr.peaks_sigma() > 2.0);
# Ok(()) }

# fn main() -> Result<(), rustyms::error::CustomError> {
use rustyms::{*, align::*};
// Check how this peptide compares to a similar peptide (using the feature `align`)
let first_peptide = LinearPeptide::pro_forma("IVQEVS", None)?.into_simple_linear().unwrap();
let second_peptide = LinearPeptide::pro_forma("LEVQVES", None)?.into_simple_linear().unwrap();
// Align the two peptides using mass based alignment
// I-VQEVS A
// LEVQVES B
// ─+  ╶─ 
let alignment = align::<4, SimpleLinear, SimpleLinear>(&first_peptide, &second_peptide,
                 matrix::BLOSUM62, Tolerance::new_ppm(10.0), AlignType::GLOBAL);
# dbg!(&alignment);
// Calculate some more statistics on this alignment
let stats = alignment.stats();
assert_eq!(stats.mass_similar, 6); // 6 out of the 7 positions are mass similar
assert_eq!(stats.gaps, 1); // 1 position is an insertion
# Ok(()) }

Compilation features

Rustyms ties together multiple smaller modules into one cohesive structure. It has multiple features which allow you to slim it down if needed (all are enabled by default).

• align - gives access to mass based alignment of peptides.
• identification - gives access to methods reading many different identified peptide formats.
• imgt - enables access to the IMGT database of antibodies germline sequences, with annotations.
• isotopes - gives access to generation of an averagine model for isotopes, also enables two additional dependencies.
• rand - allows the generation of random peptides.
• rayon - enables parallel iterators using rayon, mostly for imgt but also in consecutive align.
• mzdata - enables integration with mzdata which has more advanced raw file support.

Sources for the downloaded files

• PSI-MOD: https://github.com/HUPO-PSI/psi-mod-CV (2021-06-13 v1.031.6)
• Unimod: http://www.unimod.org/obo/unimod.obo (2024-08-12 11:33)
• RESID: ftp://ftp.proteininformationresource.org/pir_databases/other_databases/resid/ (2018-04-31 RESIDUES.XML)
• XL-MOD: https://raw.githubusercontent.com/HUPO-PSI/mzIdentML/master/cv/XLMOD.obo (2021-03-23 1.1.12)
• GNO: http://purl.obolibrary.org/obo/gno.obo (2024-05-21) structures: https://glycosmos.org/download/ ('List of all GlyCosmos Glycans data.') (downloaded 2024-07-02)
  • To save space (crates.io has a hard limit on crate size) the unused columns of the structures csv are remove (only 0 and 1 are kept) and the gno.obo is trimmed using the following regex: (property_value: GNO:00000(022|023|041|042|101|102) .*$\n)|(def: .*$\n)|(synonym: .*$\n)|(name: [^ ]*$\n) (any matching line is removed) and the following replacement regex (is_a: [^ ]*) ! .*\n with $1\n.
  • The structures csv file has only the first two columns kept for the same reason, also remove the two lines starting with "
• Isotopic atomic masses: https://ciaaw.org/data/IUPAC-atomic-masses.csv (2021-03-17)

Ontologies

Note some modifications that do not fit the assumptions of rustyms might be missing from the ontologies. Examples of these are cross-links with more then 2 positions from XL-MOD and RESID, and modifications with different diff_formulas based on which location they bound from RESID. Additionally only the Glycans of a specific mass or with a structure.