binary-classification-output-interpretation
Summary
Interpret and compile binary classification predictions (bitter/not-bitter) from a machine-learning classifier into structured output files linked to molecule identifiers. This skill bridges trained model inference and downstream analysis by organizing predictions into a reusable tabular format.
When to use
You have executed a binary classifier (such as BitterPredict.m) on a set of molecules with chemical structure descriptors and need to translate the raw predictions into a structured CSV output file that maps molecule identifiers to their predicted class labels (bitter or not-bitter).
When NOT to use
- The classifier output is already in a structured tabular format with molecule IDs and predictions — no recompilation is needed.
- You need confidence scores or probability estimates rather than binary class labels alone.
- The input molecules lack consistent identifiers that can be reliably mapped to predictions.
Inputs
- Binary classification predictions (bitter/not-bitter) from BitterPredict.m classifier
- Molecule identifiers (from input descriptor file)
- Chemical structure descriptor data (from CSV or Excel input file)
Outputs
- CSV file containing molecule identifiers and predicted class labels (bitter or not-bitter)
How to apply
After BitterPredict.m generates binary bitter/not-bitter predictions for each molecule in your input descriptor set, collect the predicted class labels alongside the corresponding molecule identifiers. Organize these pairs into a structured table with columns for molecule ID and predicted class. Write the compiled table to a CSV output file. Verify that each molecule in the input has exactly one prediction and that all class labels are valid binary values (bitter or not-bitter). The output CSV should maintain the same row order as the input molecules to ensure traceability from input descriptors through prediction to final results.
Related tools
- BitterPredict.m (Executes binary classification on molecular descriptors to generate bitter/not-bitter predictions for each molecule) — https://github.com/Niv-Lab/BitterPredict1
Evaluation signals
- Every molecule identifier from the input descriptor file appears exactly once in the output CSV.
- Every prediction in the output CSV is a valid binary class label (bitter or not-bitter); no null, missing, or invalid values.
- The row order in the output CSV matches the row order of molecules in the input descriptor file, enabling traceability.
- The output CSV is parseable as a valid CSV file with consistent column structure and no formatting errors.
- Spot-check a sample of predictions by re-running BitterPredict.m on a subset of input molecules and confirming predictions match the compiled output.
Limitations
- The skill assumes BitterPredict.m is fully executed before interpretation; if the classifier fails or produces incomplete predictions, the output CSV will be incomplete or incorrect.
- Full code availability is contingent on publication; current repository access may be restricted and users should contact nivlab@agri.huji.ac.il for availability.
- The CSV output contains only binary class labels; probability scores or confidence metrics from the classifier are not captured.
- Molecule identifiers must be present and consistent across input and output; missing or mismatched IDs will break traceability.
Evidence
- [readme] BitterPredict.m gets as input CSV or EXCEL files with required descriptors of molecules, and calucautes a predictions if each molecule is bitter or not.: "BitterPredict.m gets as input CSV or EXCEL files with required descriptors of molecules, and calucautes a predictions if each molecule is bitter or not."
- [other] Generate binary predictions (bitter or not-bitter) for each molecule. 4. Compile predictions into output CSV file with molecule identifiers and predicted class labels.: "Generate binary predictions (bitter or not-bitter) for each molecule. 4. Compile predictions into output CSV file with molecule identifiers and predicted class labels."
- [readme] BitterPredict is a classifier which predicts whether a compound is bitter or not, based on its chemical structure.: "BitterPredict is a classifier which predicts whether a compound is bitter or not, based on its chemical structure."