# rathanpaida/drug_discovery_research architecture

> Architecture diagram of the GitHub repository rathanpaida/drug_discovery_research, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-11-20). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/rathanpaida/drug_discovery_research
- Repository: https://github.com/rathanpaida/drug_discovery_research

## Overview

TNK1 is a research platform for predicting kinase activity and generating, evaluating, and prioritizing candidate molecules. The documented workflow runs from pan-kinome data and molecular features through transfer learning, uncertainty calibration, profiling, reinforcement-learning generation, novelty checks, and docking toward a wet-lab contract. The graph uses the root package as the canonical implementation; the tree also contains a parallel TO_RUN_ON_GPU copy. Workflow edges rely on README documentation, while internal model and scorer links use the provided source index and excerpts.

Main entry points: [`tnk1/models/pipeline.py`](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/models/pipeline.py) (Transfer Training), [`tnk1/rl/scorer.py`](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/rl/scorer.py) (Candidate Scoring), [`rl_min/run_rl.py`](https://github.com/rathanpaida/drug_discovery_research/blob/main/rl_min/run_rl.py) (Policy Training), [`tnk1/baselines/features.py`](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/baselines/features.py) (Molecular Features)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_data["Data and Features"]
  node_activity["Activity Data<br/>[activity.py]"]
  node_splits["Data Splits<br/>[splits.py]"]
  node_features["Molecular Features<br/>[features.py]"]
  node_context["Stage Context<br/>[context.py]"]
end

subgraph group_predict["Modeling and Uncertainty"]
  node_modeldata["Model Data<br/>[data.py]"]
  node_kinasefeatures["Kinase Features<br/>[kinase_features.py]"]
  node_training["Transfer Training<br/>[pipeline.py]"]
  node_model["Kinome Model<br/>[model.py]"]
  node_trainer["Training Engine<br/>[trainer.py]"]
  node_ensemble["Model Ensemble<br/>[ensemble.py]"]
  node_uq["Uncertainty Metrics<br/>[metrics.py]"]
  node_calibration["Conformal Calibration<br/>[calibration.py]"]
end

subgraph group_candidate["Candidate Evaluation"]
  node_scorer["Candidate Scoring<br/>[scorer.py]"]
  node_selectivity["Selectivity Profile<br/>[selectivity.py]"]
  node_admet["ADMET Profile<br/>[admet.py]"]
  node_filters["Candidate Filters<br/>[filters.py]"]
  node_novelty["Novelty Audit<br/>[verifier.py]"]
  node_docking["Molecular Docking"]
  node_contract["Wet-Lab Contract<br/>[contract.py]"]
end

subgraph group_generation["Molecule Generation"]
  node_policy["Policy Training<br/>[run_rl.py]"]
  node_dap["DAP Algorithm<br/>[dap.py]"]
  node_policy_model["Policy Model<br/>[model.py]"]
  node_vocab["SMILES Vocabulary<br/>[vocab.py]"]
end

node_researcher(("Researcher"))

node_researcher -.->|"provides data"| node_activity
node_activity -->|"feeds features"| node_features
node_features -->|"supports training"| node_training
node_splits -->|"partitions data"| node_training
node_context -->|"loads features"| node_kinasefeatures
node_context -->|"loads descriptors"| node_features
node_training -->|"builds datasets"| node_modeldata
node_training -->|"creates model"| node_model
node_training -->|"runs epochs"| node_trainer
node_training -->|"uses features"| node_kinasefeatures
node_training -->|"reports metrics"| node_uq
node_training -->|"uses folds"| node_splits
node_training -->|"feeds ensemble"| node_ensemble
node_ensemble -->|"calibrates predictions"| node_calibration
node_calibration -->|"informs scoring"| node_scorer
node_scorer -->|"uses profile"| node_selectivity
node_scorer -->|"uses profile"| node_admet
node_scorer -->|"applies filters"| node_filters
node_scorer -->|"scores candidates"| node_policy
node_policy -->|"trains policy"| node_dap
node_policy -->|"uses model"| node_policy_model
node_policy -->|"uses vocabulary"| node_vocab
node_policy -->|"generated molecules audited"| node_novelty
node_novelty -->|"passes candidates"| node_docking
node_docking -->|"supports prioritization"| node_contract

click node_activity "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/data/activity.py"
click node_splits "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/data/splits.py"
click node_features "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/baselines/features.py"
click node_context "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/context.py"
click node_modeldata "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/models/data.py"
click node_kinasefeatures "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/models/kinase_features.py"
click node_training "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/models/pipeline.py"
click node_model "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/models/model.py"
click node_trainer "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/models/trainer.py"
click node_ensemble "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/uq/ensemble.py"
click node_uq "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/uq/metrics.py"
click node_calibration "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/uq/calibration.py"
click node_scorer "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/rl/scorer.py"
click node_selectivity "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/selectivity/selectivity.py"
click node_admet "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/admet/admet.py"
click node_filters "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/admet/filters.py"
click node_novelty "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/novelty/verifier.py"
click node_docking "https://github.com/rathanpaida/drug_discovery_research/blob/main/scripts/dock_40_molecules_tnk1.py"
click node_contract "https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/active_learning/contract.py"
click node_policy "https://github.com/rathanpaida/drug_discovery_research/blob/main/rl_min/run_rl.py"
click node_dap "https://github.com/rathanpaida/drug_discovery_research/blob/main/rl_min/dap.py"
click node_policy_model "https://github.com/rathanpaida/drug_discovery_research/blob/main/rl_min/model.py"
click node_vocab "https://github.com/rathanpaida/drug_discovery_research/blob/main/rl_min/vocab.py"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_activity,node_splits,node_features,node_context toneBlue
class node_modeldata,node_kinasefeatures,node_training,node_model,node_trainer,node_ensemble,node_uq,node_calibration toneAmber
class node_scorer,node_selectivity,node_admet,node_filters,node_novelty,node_docking,node_contract toneMint
class node_policy,node_dap,node_policy_model,node_vocab toneRose
class node_researcher toneIndigo
```

## Components

### Data and Features

- [Activity Data](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/data/activity.py): `tnk1/data/activity.py`
- [Data Splits](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/data/splits.py): `tnk1/data/splits.py`
- [Molecular Features](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/baselines/features.py): `tnk1/baselines/features.py`
- [Stage Context](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/context.py): `tnk1/context.py`

### Modeling and Uncertainty

- [Model Data](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/models/data.py): `tnk1/models/data.py`
- [Kinase Features](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/models/kinase_features.py): `tnk1/models/kinase_features.py`
- [Transfer Training](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/models/pipeline.py): `tnk1/models/pipeline.py`
- [Kinome Model](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/models/model.py): `tnk1/models/model.py`
- [Training Engine](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/models/trainer.py): `tnk1/models/trainer.py`
- [Model Ensemble](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/uq/ensemble.py): `tnk1/uq/ensemble.py`
- [Uncertainty Metrics](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/uq/metrics.py): `tnk1/uq/metrics.py`
- [Conformal Calibration](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/uq/calibration.py): `tnk1/uq/calibration.py`

### Candidate Evaluation

- [Candidate Scoring](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/rl/scorer.py): `tnk1/rl/scorer.py`
- [Selectivity Profile](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/selectivity/selectivity.py): `tnk1/selectivity/selectivity.py`
- [ADMET Profile](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/admet/admet.py): `tnk1/admet/admet.py`
- [Candidate Filters](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/admet/filters.py): `tnk1/admet/filters.py`
- [Novelty Audit](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/novelty/verifier.py): `tnk1/novelty/verifier.py`
- [Molecular Docking](https://github.com/rathanpaida/drug_discovery_research/blob/main/scripts/dock_40_molecules_tnk1.py): `scripts/dock_40_molecules_tnk1.py`
- [Wet-Lab Contract](https://github.com/rathanpaida/drug_discovery_research/blob/main/tnk1/active_learning/contract.py): `tnk1/active_learning/contract.py`

### Molecule Generation

- [Policy Training](https://github.com/rathanpaida/drug_discovery_research/blob/main/rl_min/run_rl.py): `rl_min/run_rl.py`
- [DAP Algorithm](https://github.com/rathanpaida/drug_discovery_research/blob/main/rl_min/dap.py): `rl_min/dap.py`
- [Policy Model](https://github.com/rathanpaida/drug_discovery_research/blob/main/rl_min/model.py): `rl_min/model.py`
- [SMILES Vocabulary](https://github.com/rathanpaida/drug_discovery_research/blob/main/rl_min/vocab.py): `rl_min/vocab.py`

### Other components

- **Researcher**

## Connections

- Researcher → Activity Data: provides data (evidence: `README.md`)
- Activity Data → Molecular Features: feeds features (evidence: `README.md`)
- Molecular Features → Transfer Training: supports training (evidence: `README.md`)
- Data Splits → Transfer Training: partitions data (evidence: `tnk1/models/pipeline.py`)
- Stage Context → Kinase Features: loads features (evidence: `tnk1/context.py`)
- Stage Context → Molecular Features: loads descriptors (evidence: `tnk1/context.py`)
- Transfer Training → Model Data: builds datasets (evidence: `tnk1/models/pipeline.py`)
- Transfer Training → Kinome Model: creates model (evidence: `tnk1/models/pipeline.py`)
- Transfer Training → Training Engine: runs epochs (evidence: `tnk1/models/pipeline.py`)
- Transfer Training → Kinase Features: uses features (evidence: `tnk1/models/pipeline.py`)
- Transfer Training → Uncertainty Metrics: reports metrics (evidence: `tnk1/models/pipeline.py`)
- Transfer Training → Data Splits: uses folds (evidence: `tnk1/models/pipeline.py`)
- Transfer Training → Model Ensemble: feeds ensemble (evidence: `README.md`)
- Model Ensemble → Conformal Calibration: calibrates predictions (evidence: `README.md`)
- Conformal Calibration → Candidate Scoring: informs scoring (evidence: `README.md`)
- Candidate Scoring → Selectivity Profile: uses profile (evidence: `tnk1/rl/scorer.py`)
- Candidate Scoring → ADMET Profile: uses profile (evidence: `tnk1/rl/scorer.py`)
- Candidate Scoring → Candidate Filters: applies filters (evidence: `tnk1/rl/scorer.py`)
- Candidate Scoring → Policy Training: scores candidates (evidence: `README.md`)
- Policy Training → DAP Algorithm: trains policy (evidence: `rl_min/run_rl.py`)
- Policy Training → Policy Model: uses model (evidence: `rl_min/run_rl.py`)
- Policy Training → SMILES Vocabulary: uses vocabulary (evidence: `rl_min/run_rl.py`)
- Policy Training → Novelty Audit: generated molecules audited (evidence: `README.md`)
- Novelty Audit → Molecular Docking: passes candidates (evidence: `README.md`)
- Molecular Docking → Wet-Lab Contract: supports prioritization (evidence: `README.md`)
