Virtual Multicomponent Crystal Screening: Hydrogen Bonding Revisited
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Pharmaceutical cocrystals, salts, and multicomponent crystals, in general, have increasingly come under the spotlight in recent years. A fast and efficient a priori theoretical classifier to identify potential coformers is highly sought after to complement the experimental brute force screening methods. This research examines the qualitative approaches that are based on hydrogen bonding strength. First, molecular electrostatic potential (MEP) maps of 330 coformers were obtained from density functional theory simulations, using two geometries: experimentally determined crystal structures and gas-phase optimization. An in-depth comparison of MEPs revealed the potential pitfalls of these two geometries that are deliberated at length in the manuscript. Next, six APIs and their reported salts/cocrystals on the Cambridge Structural Database (CSD) were inversely predicted with MEP analysis. For two of these APIs, the prediction showed systematic errors that are resolved with suggestions provided in the manuscript. Subsequently, hydrogen bond energy (HBE) and hydrogen bond propensity (HBP) calculations were put to the test with two APIs and 52 organic coformers. Finally, multivariate logistic regression, a linear machine learning (ML) algorithm, showed how a combination of HBE and HBP can be a superior classifier, for which 18 out of 25 positive cases were uninterruptedly identified at the top of the list. Provided that a database of failed attempts of cocrystallization is compiled within the scientific community to supplement the existing positive results (multicomponent crystals in the CSD), the combination of chemistry-based parameters and ML can be a promising classifier for coformer selection.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it