Iterative Fragment Selection: A Group Contribution Approach to Predicting Fish Biotransformation Half-Lives
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
There are regulatory needs to evaluate thousands of chemicals for potential hazard and risk with limited available information. An automated method is presented for developing and evaluating Quantitative Structure-Activity Relationships (QSARs) for a range of chemical properties that can be applied for screening level chemical assessments. The method is an integrated algorithm for descriptor generation, data set splitting, cross validation, and model selection. Resulting QSARs are two-dimensional (2D) fragment-based group contribution models. The QSAR development and evaluation method does not require previous expert knowledge for selecting 2D fragments associated with the chemical property of interest. The method includes information on the domain of applicability (structural similarity to the training set) and estimates of the uncertainty in the QSAR predictions. As a demonstration, the method is applied to generate novel QSARs for fish primary biotransformation half-lives (HL(N)). Results from the new HL(N) QSARs are compared to another 2D fragment-based HL(N) QSAR developed with expert judgment, and the predictive powers of the models are found to be similar. The relative merits and limitations of each method are investigated and the new QSAR is found to make comparable predictions with significantly fewer fragments. A coefficient of determination (R(2)) of 0.789 and a root mean squared error (RMSE) of 0.526 were obtained for the training data set and an R(2) of 0.748 and an RMSE of 0.584 were obtained for the validation data set, along with a concordance correlation coefficient (CCC) of 0.857 showing good predictive power.
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 | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.000 | 0.002 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 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