Direct Steering of de novo Molecular Generation using Descriptor Conditional Recurrent Neural Networks (cRNNs)
Bibliographic record
Abstract
Deep learning has acquired considerable momentum over the past couple of years in the domain of de-novo drug design. Particularly, transfer and reinforcement learning have demonstrated the capability of steering the generative process towards chemical regions of interest. In this work, we propose a simple approach to the focused generative task by constructing a conditional recurrent neural network (cRNN). For this purpose, we aggregate selected molecular descriptors along with a QSAR-based bioactivity label and transform them into initial LSTM states before starting the generation of SMILES strings that are focused towards the aspired properties. We thus tackle the inverse QSAR problem directly by training on molecular descriptors, instead of iteratively optimizing around a set of candidate molecules. The trained cRNNs are able to generate molecules near multiple specified conditions, while maintaining an output that is more focused than traditional RNNs yet less focused than autoencoders. The method shows promise for applications in both scaffold hoping and ligand series generation, depending on whether the cRNN is trained on calculated scalar molecular properties or structural fingerprints. This also demonstrates that fingerprint-to-molecule decoding is feasible, leading to molecules that are similar – if not identical – to the ones the fingerprints originated from. Additionally, the cRNN is able to generate a larger fraction of predicted active compounds against the DRD2 receptor when compared to an RNN trained with the transfer learning model.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".