Predictive Models for Crystallization-Induced Diastereomer Transformations

A 786.4 KB dataset from figshare contains features and models for predicting diastereoselectivity in crystallization-induced diastereomer transformations (CIDTs). The workflow by Sara L. McCormack featurizes product structures to build statistical models, validated on six previously untested substrates. Feature analysis identifies amide identity, conformational compactness, and local electronic properties as key determinants of stereochemical outcome.

Use Cases

• Predicting diastereoselectivity outcomes in CIDT experiments based on structural features.
• Training machine learning models for organic synthesis planning using identified key determinants like amide identity.
• Analyzing structure-selectivity relationships for labile stereogenic centers based on conformational compactness and electronic properties.

Strengths

• Model validation on six previously untested substrates is explicitly mentioned.
• Feature analysis provides a mechanistic basis by highlighting specific molecular properties.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset's 786.4 KB size indicates a limited scope of data.

Provenance

Source

Sara L. McCormack via figshare

Collection Method

Workflow to featurize product structures and construct statistical models, as described.

Freshness

Last updated 2026-06-04 17:07:29; freshness should be verified.