Prerequisite: Load The Molecule From ChemrytIQ

Before opening ChemrytNMR, search the molecule in ChemrytIQ by SMILES, InChI, molecule name, or CAS number. Confirm the correct molecule on the ChemrytIQ page, then open the required Chemryt app from that same molecule context so the selected structure is loaded into the app automatically.

What ChemrytNMR Does

ChemrytNMR supports single- and multi-molecule NMR prediction, 1H and heteronucleus workflows, exact or first-order spin treatment, conformer and 2D experiments, experimental-spectrum overlay, processing and deconvolution, quantitative NMR, atom-to-peak mapping, and exportable reports.

Prediction & Simulation

Predict 1H, 13C, 15N, 19F, or 31P spectra with acquisition, reference, spin, conformer, and 2D experiment settings.

Verification & Overlay

Import JCAMP-DX, CSV, or zipped Bruker data and compare experimental and predicted spectra using configurable alignment and scoring.

Quantitative & Processing

Configure zero filling, apodization, phase and baseline correction, peak picking, deconvolution, relaxation-aware intensity, and qNMR.

Quick Workflow

Choose `Single Molecule` for one structure or `Multi Molecule` for a sample containing several known components.
Draw in JSME, paste SMILES, load a ChemrytIQ structure, or upload MOL, SDF, MOL2, or XYZ coordinates. Confirm formula, InChIKey, stereochemistry, and the authoritative SMILES.
In `Prediction & Simulation`, select nucleus, 13C acquisition mode when relevant, shift reference and offset, solvent, scale, field, pH, line width, simulation precision, 1H spin treatment, and optional 2D experiment.
Click `Predict NMR` and review the spectrum, predicted shifts, J-couplings, atom mapping, conformer summary, 2D map, interpretation, and warnings.
Use `Verification & Overlay` to import an experimental spectrum, choose tolerance, normalization, match metric, threshold, and display mode, then run the comparison.
Use `Quantitative & Processing` to configure processing, peak picking, deconvolution, relaxation-aware intensity, and qNMR standard or region settings.
Export the report only after confirming that the structure, acquisition context, experimental file, processing parameters, and current results belong to the same analysis.

Main Areas

Area	What to enter or review	When to use it
Sample setup	Mode switch, JSME editor, SMILES import, ChemrytIQ handoff, 3D structure upload, sample builder, ratios, selected sample list, and identity summary.	Use to define exactly what the predicted or combined spectrum represents.
Single Molecule vs Multi Molecule workflow	Single Molecule uses one drawn or pasted structure as the prediction subject. Multi Molecule opens the sample builder so users can add several structures, label them, set component ratios, enable or disable entries, bulk import molecules, duplicate entries, and load a demo mixture.	Use Single Molecule for assignment of one compound, atom-to-peak review, 2D maps, and clean structure verification. Use Multi Molecule for mixture, formulation, impurity, solvent, coformer, or reaction-sample screening where the displayed spectrum should represent more than one component.
Prediction & Simulation	Nucleus; coupled or decoupled 13C; TMS, CFCl3, H3PO4, CH3NO2, or custom reference; offset; solvent; scale; field; pH; line width; precision; spin treatment; 2D experiment; NOE mixing time.	Use to match the prediction and simulation to the intended experiment.
2D Experiments	COSY, HSQC, HSQC-TOCSY, HMBC, NOESY, and ROESY where supported. Uploaded 3D coordinates can provide explicit NOESY/ROESY distance context.	Select before prediction when cross-peak maps are needed for connectivity or spatial interpretation.
Verification & Overlay	Experimental JCAMP-DX, CSV, or zipped Bruker file; ppm tolerance; maximum, integral, or RMS normalization; cosine, DP4, or RMSE metric; match threshold; overlay, stacked, or mirror view.	Use to compare a proposed structure prediction with measured data.
Quantitative & Processing	Zero filling, exponential/Gaussian/sine-bell apodization, phase correction, baseline correction, SNR threshold, d1, flip angle, T1, intensity model, qNMR standard, and deconvolution tolerance.	Use to process the current spectrum and configure quantitative interpretation.
qNMR setup	Predicted profile or peak table; per-signal, manual, or full-spectrum regions; standard and target peaks; proton counts; concentration; molecular weights.	Use after choosing well-resolved regions and documenting the reference-standard basis.
Results and analysis toolbar	Spectrum, literature reference toggle, predicted shifts, J-couplings, Peak Pick, qNMR, Deconvolution, molecule-wise breakdown, overlay actions, and report export.	Use to review assignments, diagnose overlap, quantify regions, and document the run.

Tutorial Notes

Start with a clean single-molecule 1H or 13C run and confirm identity before moving to mixtures, 2D experiments, verification, or qNMR.
Use atom hover, highlighted structure, predicted-shift table, J-coupling table, and spectrum together; a peak position alone is not a validated assignment.
For 13C, choose proton-decoupled 13C{1H} for normal singlet-style prediction or proton-coupled 13C when first-order carbon-proton splitting is intentionally required.
Use the reference standard appropriate to the nucleus: TMS is the default, while CFCl3, 85% H3PO4, and CH3NO2 support 19F, 31P, and 15N contexts. Use Custom only with a documented offset.
For 1H, `Auto exact QM` is intended to model strong coupling when the spin system requires it; `First-order` provides a simpler splitting representation. Field strength changes Hz separation and multiplet appearance, not ppm shift positions.
Simulation precision controls computation depth: Fast GNN for routine screening, Multiplets only for splitting emphasis, Conformer ensemble for flexible molecules and J averaging, and Simple shifts when only predicted centers are required.
COSY supports short-range proton coupling, HSQC one-bond H-C correlations, HSQC-TOCSY relay connectivity, HMBC longer-range H-C correlations, and NOESY/ROESY spatial proximity. Enter a NOE mixing time when those modes are selected.
Upload MOL, SDF, MOL2, or XYZ coordinates when known 3D geometry should drive NOESY/ROESY distances. Verify that atom ordering and structure identity match the prediction subject.
In Multi Molecule mode, label each component, choose parts or percentage, set a positive amount, and enable only contributing molecules. The combined trace is a weighted simulated sample, not an automatic concentration measurement.
Verification accepts JCAMP-DX, CSV, or zipped Bruker datasets. Confirm ppm direction, nucleus, reference, field, and preprocessing before interpreting cosine, DP4, or RMSE scores.
Choose overlay normalization deliberately: maximum peak compares shapes, total integral compares area-normalized spectra, and RMS normalization emphasizes overall pointwise magnitude. Inspect the graphical overlay as well as the score.
For processing, record zero filling, window function, line broadening, phase method, baseline method, and SNR threshold. Changing processing can change apparent peaks, areas, and fit quality.
For relaxation-aware qNMR, enter d1, flip angle, and measured or justified T1. Define a reliable standard and target region with known proton counts, concentrations, and molecular weights.
Peak Pick reports detected theoretical centers and assignments; Deconvolution separates overlapped predicted envelopes using the selected tolerance. Treat fitted components as a model until confirmed against the raw experimental data.
The toolbar exposes only actions appropriate to the active workflow and available data. Overlay and ASV remain disabled until an experimental spectrum is loaded.
Export Report captures the current graph, data, methods, molecule information, and disclaimer. Rerun the workflow after any structure or setting change before exporting.

Complete Single-Molecule Prediction

Use this workflow to predict and assign one structure.

Draw the molecule or paste SMILES, click `Load Structure`, and verify formula, InChIKey, stereochemistry, and current SMILES.
Open `Prediction & Simulation`; choose nucleus, acquisition and reference settings, solvent, field, pH, line width, simulation precision, spin treatment, and optional 2D experiment.
Click `Predict NMR` and wait for the progress state to finish.
Inspect the spectrum and Predicted Shifts tab, then use atom hover or table interactions to verify structure-to-signal mapping.
Review J-Couplings for 1H or coupled 13C, Conformer Averaging when requested, and the 2D map when an experiment was selected.
Use Peak Pick or Deconvolution for crowded regions and export only after checking warnings and method metadata.

Experimental Verification & Overlay

Use this workflow to compare a proposed structure with measured data.

Run a prediction for the correct nucleus and acquisition context first.
Open `Verification & Overlay` and import JCAMP-DX, CSV, or a zipped Bruker dataset.
Confirm file parsing, experimental metadata, ppm axis, reference, nucleus, and point count.
Set peak tolerance, normalization, score metric, threshold, and overlay/stacked/mirror display.
Click `Run Overlay & Match`, inspect aligned traces and unmatched or shifted regions, and use the selected score as supporting—not standalone—evidence.
Record solvent, concentration, temperature, field, pulse sequence, processing, and any reason for mismatch.

Quantitative & Processing Workflow

Use this workflow for controlled processing, peak extraction, fitting, and qNMR setup.

Open `Quantitative & Processing` and set zero filling, apodization and width, phase correction, baseline correction, SNR, and deconvolution tolerance.
For quantitative work, enter d1, flip angle, measured/default T1, intensity model, and qNMR standard type.
Click `Run Processing`, then review Peak Pick and Deconvolution results for artifacts, unresolved overlap, and baseline sensitivity.
Open qNMR, choose predicted profile or peak table, select per-signal/manual/full regions, and define the standard and target peaks.
Enter standard and target proton counts, standard concentration, and molecular weights when required, then click `Run qNMR`.
Validate quantitative output with measured spectrum quality, complete relaxation, calibrated response, weighed standards, purity corrections, and uncertainty assessment.

Multi-Molecule Sample Workflow

Use this workflow for known mixtures, coformers, reaction samples, or impurity-spiking simulations.

Switch to `Multi Molecule`, draw or paste the first component, give it a label, choose parts or percentage, and enter its relative amount.
Click `Add To Sample`; repeat, use Bulk Import SMILES, or load the demo sample. Edit, duplicate, enable, or disable entries as needed.
Check that enabled ratios are positive and the intended sample total is represented before prediction.
Run prediction and review the Combined Sample Spectrum, component overlays, overlap clusters, interpretation, and molecule-wise breakdown.
Return to Single Molecule for detailed assignment, 2D review, or deconvolution of a component responsible for a crowded signal.

Troubleshooting

Problem	Likely cause	What to do
Prediction does not start	No valid structure is loaded, the sample list has no enabled positive-ratio component, or the backend is unavailable.	Validate the structure/sample, use Check Server if exposed, read the workflow status, and retry after correcting the reported issue.
Experimental overlay button is disabled	No supported experimental spectrum has been loaded successfully.	Import .dx, .jdx, .csv, or a zipped Bruker dataset and wait for the file status to confirm parsing.
Overlay is shifted or scores poorly	Prediction and experiment may use different referencing, solvent, nucleus, or acquisition/processing conditions.	Align the reference and ppm axis, verify metadata, adjust tolerance only with justification, and inspect the traces rather than optimizing the score alone.
No 2D map appears	The selected nucleus/experiment is unsupported, the structure lacks the required correlations, or prediction was not rerun.	Choose a supported combination, rerun prediction, and read backend cross-peak warnings.
NOESY/ROESY distances look wrong	The conformer or uploaded 3D file may not represent the experimental geometry or atom order.	Verify the MOL/SDF/MOL2/XYZ structure and conformer, then treat proximity output as a hypothesis until experimentally confirmed.
qNMR result is unrealistic	Regions overlap, baseline/phase is poor, relaxation is incomplete, or standard/proton-count/concentration inputs are wrong.	Choose isolated regions, correct processing and metadata, use adequate d1, confirm T1 and standard values, and rerun.
Peak count changes after processing	SNR, zero filling, apodization, line width, phase, baseline, or deconvolution tolerance changed.	Document processing parameters and verify each new peak against the raw spectrum and assignment context.
Report contains outdated results	The structure or settings changed after the last completed workflow.	Rerun prediction/verification/processing, confirm the latest timestamp and panels, then export again.

ML Model / Computation Used

Model or method	What it predicts	Implementation details
ChemrytNMR atom GNN shift models	Atom-level chemical shifts for 1H, 13C, 15N, 19F, and 31P workflows.	PyTorch GNN artifacts are stored under chemrytnmr_models. The 1H model reports validation MAE about 0.31 ppm; the 13C model reports validation MAE about 3.06 ppm on nmrshiftdb2-derived labels.
NMR processing and workflow models	Peak picking, overlays, conformer/2D support, qNMR, and report calculations.	These steps combine Python-backed processing, rule/parameter logic, and spectrum post-processing around the GNN shift predictions.

Good Practice

ChemrytNMR predictions, overlays, fitted components, 2D maps, and qNMR outputs are analytical decision support. Confirm assignments and quantitative conclusions with properly referenced measured spectra, validated acquisition and processing, suitable standards, uncertainty checks, and orthogonal analytical evidence.

Reference Used

This Tutorial page was prepared from the ChemrytLabs reference module: ChemrytNMR.