chem-nmr-analysis

Quantify component mole fractions in a 1H NMR mixture spectrum by deconvolving the mixture spectrum against known reference spectra using Wasserstein-distance optimal transport. Supply per-component proton counts to convert the optimal transport weights into accurate mole fractions.

Wasserstein distance deconvolution works for NMR spectral analysis by treating spectral intensity distributions as probability measures and using optimal transport to find the best mixture weights. This approach matches mixture spectra against reference components to estimate proportions and generate fit-quality metrics.

Predict reaction products from reactant or reagent SMILES via ReactionT5 to generate hypothesized components. These predicted products can supplement your known reference spectra for Wasserstein deconvolution when analyzing crude reaction mixtures.

Analyze reaction kinetics from time-series NMR spectra by performing supervised deconvolution across multiple crude spectra time points. Apply Wasserstein fitting to each time point to produce a kinetics table and overlay plots for end-to-end reaction monitoring.

Provide mixture and reference spectra as two-column ppm and intensity files for NMR mixture deconvolution. Optionally apply baseline correction to the input files and supply per-component proton counts to convert spectral proportions into mole fractions.

Optimal transport for NMR spectral deconvolution requires known or hypothesized component identities and corresponding reference spectra. Fit quality depends on reference accuracy, and the method outputs machine-readable JSON with metrics to help evaluate deconvolution limitations and spectral matching performance.