Primary sequence based protein–protein interaction binder generation with transformers
Bibliographic record
Abstract
Abstract The design of binder proteins for specific target proteins using deep learning is a challenging task that has a wide range of applications in both designing therapeutic antibodies and creating new drugs. Machine learning-based solutions, as opposed to laboratory design, streamline the design process and enable the design of new proteins that may be required to address new and orphan diseases. Most techniques proposed in the literature necessitate either domain knowledge or some appraisal of the target protein’s 3-D structure. This paper proposes an approach for designing binder proteins based solely on the amino acid sequence of the target protein and without recourse to domain knowledge or structural information. The sequences of the binders are generated with two new transformers, namely the AppendFormer and MergeFormer architectures. Because, in general, there is more than one binder for a given target protein, these transformers employ a binding score and a prior on the sequence of the binder to obtain a unique targeted solution. Our experimental evaluation confirms the strengths of this novel approach. The performance of the models was determined with 5-fold cross-validation and clearly indicates that our architectures lead to highly accurate results. In addition, scores of up to 0.98 were achieved in terms of Needleman-Wunsch and Smith-Waterman similarity metrics, which indicates that our solutions significantly outperform a seq2seq baseline 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 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
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".