PyDSSP-rs

A Rust implementation of the DSSP algorithm, ported from PyDSSP.

This project provides a fast and efficient way to assign secondary structures to protein backbone coordinates using the DSSP algorithm. It utilizes the ndarray crate for numerical operations.

Features

• DSSP Algorithm: Implements the core logic for hydrogen bond energy calculation and secondary structure assignment (Helix, Strand, Loop).
• NDArray Backend: Uses ndarray for efficient array manipulations.
• CLI Tool: Includes a command-line interface dssp for processing PDB files.
• High Correlation: Verified against the original DSSP implementation with >97% correlation on the TS50 dataset.

Installation

Ensure you have Rust installed.

git clone https://github.com/mogura-rs/PyDSSP-rs.git
cd PyDSSP-rs
cargo build --release

The binary will be available at target/release/dssp.

Usage

Command Line Interface

You can run the dssp tool to process PDB files.

# Process a single file and output to stdout
cargo run --bin dssp -- input.pdb

# Process multiple files
cargo run --bin dssp -- input1.pdb input2.pdb

# Save output to a file
cargo run --bin dssp -- input.pdb -o output.result

Library

Add pydssp-rs to your Cargo.toml.

[dependencies]
pydssp-rs = { git = "https://github.com/mogura-rs/PyDSSP-rs" }
ndarray = "0.17.2"

use pydssp_rs::{read_pdbtext, assign};
use ndarray::Axis;

fn main() {
    let pdb_content = std::fs::read_to_string("input.pdb").unwrap();
    let (coord, sequence) = read_pdbtext(&pdb_content);

    // coord is Array3 (L, 4, 3) [N, CA, C, O]
    // sequence is String (1-letter code)

    // Create donor mask (exclude Proline)
    let l = coord.len_of(Axis(0));
    let mut donor_mask = ndarray::Array1::<f64>::ones(l);
    for (i, c) in sequence.chars().enumerate() {
        if c == 'P' { donor_mask[i] = 0.0; }
    }

    // Assign secondary structure
    // Returns Array2 (L, 3) where [Loop, Helix, Strand] are 0/1
    let result = assign(&coord.view(), Some(&donor_mask));

    println!("{:?}", result);
}

License

MIT