Abstract B009: missense-kinase-toolkit: A toolkit to facilitate kinase sequence and structure-based modeling for property predictions
Bibliographic record
Abstract
Abstract Kinases are a family of proteins that catalyze phosphorylation or the addition of an ATP-derived phosphate group to substrate proteins, lipids, or carbohydrates, which in turn potentiates numerous intracellular signaling cascades. Given the centrality of phosphorylation in growth, proliferation, motility, differentiation, and other essential biological processes, they are among the most frequently targeted proteins in drug discovery with more than 80 small molecules approved in by the U.S. FDA since 1999. Various databases provide kinase-specific sequence, structural, biochemical assay, and clinical alteration data but standardizing these resources poses a non-trivial challenge that requires familiarity with kinases’ unique properties. Here, we introduce an open-sourced, Python-based toolkit, which integrates and harmonizes a variety of resources to produce a single repository of human protein kinase information to facilitate the development of sequence and structure-based machine learning property prediction models. We use this toolkit to finetune ESM-2, a pretrained, transformer-based, masked protein language model, to predict the ATP affinity of both wild-type and mutant kinases in a low-data regime given several structurally and functionally significant sequence representations as inputs. We in turn use this model to generate predictions of the impact on ATP affinity of over 4,000 missense kinase mutations observed in a cohort of cancer patients screened using the MSK-IMPACT gene panel. Finally, we compare our findings to results from other more coarse-grained variant effect prediction algorithms, including the AF2-derived AlphaMissense. This provides a valuable use case for the utility of a well-annotated kinase database in facilitating the development of machine learning models for protein property predictions that can provide insight into the druggability of missense mutant kinases. Citation Format: Jessica White, Wesley Tansey. missense-kinase-toolkit: A toolkit to facilitate kinase sequence and structure-based modeling for property predictions [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Artificial Intelligence and Machine Learning; 2025 Jul 10-12; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(13_Suppl):Abstract nr B009.
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 | 0.002 | 0.004 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.004 | 0.002 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.022 | 0.009 |
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".