Crate qsar 

qsar — a small, focused Rust library for QSAR-related utilities.

This crate provides:

• descriptors — a lightweight exact molecular-weight calculator from SMILES/InChI (small, pure-Rust parser).
• data_io — small CSV helpers to load descriptor matrices and target vectors.
• models — helpers to convert descriptor vectors into ndarray and a minimal Linfa example.

The initial release intentionally avoids heavy native FFI dependencies (Open Babel / RDKit) so the crate stays portable and easy to publish. Full parser/toolkit integration can be added later behind optional Cargo features.

Quick examples

Compute exact molecular weight (SMILES):

use qsar::molecular_weight;

let mw = qsar::descriptors::molecular_weight("CCO").expect("compute MW");
println!("Ethanol MW: {}", mw);

Convert vectors to ndarray and train a simple linear model (see models):

use qsar::models::train_and_predict_example;
let pred = train_and_predict_example().expect("train and predict");
println!("Prediction: {:?}", pred);

Load descriptors from CSV (see data_io):

use qsar::data_io::read_csv_descriptors;
let (x, y) = read_csv_descriptors("data/dataset.csv", &["mol_wt", "logp"], "pIC50")?;

Re-exports

pub use descriptors::molecular_weight;

pub use data_io::read_csv_descriptors;

pub use models::to_ndarrays;

pub use models::train_and_predict_example;

Modules

data_io

CSV/JSON data I/O helpers for QSAR descriptor datasets.

descriptors

Molecular descriptor calculations.

models

Utilities to convert descriptor vectors into ndarray and a minimal Linfa example.