# nirmalkumar31/rbp-protocol-calibration architecture

> Architecture diagram of the GitHub repository nirmalkumar31/rbp-protocol-calibration, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-02). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/nirmalkumar31/rbp-protocol-calibration
- Repository: https://github.com/nirmalkumar31/rbp-protocol-calibration
- Video tour (about a minute): https://gitdiagram.com/nirmalkumar31/rbp-protocol-calibration/video

## Overview

This repository is a research pipeline for measuring how negative-example construction changes apparent RNA-binding-protein sequence-model performance, with an earlier variant-scoring study retained alongside it. The main workflow assembles ENCODE peak data and sequence windows, constructs negative sets and chromosome-blocked folds, trains models and composition baselines, then estimates and compares their contributions. Cloud scripts support ingestion and GPU sweeps; local analysis assembles evidence, while verification checks published values. The graph combines sampled source references with README-documented capabilities; unsampled implementation details are not inferred.

Main entry points: [`scripts/cloud_analysis.py`](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/scripts/cloud_analysis.py) (Cloud analysis), [`src/rbp/models/registry.py`](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/models/registry.py) (Model registry), [`src/rbp/data/windows.py`](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/data/windows.py) (Sequence windows), [`src/rbp/train/trainer.py`](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/train/trainer.py) (Model trainer)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_data["Data preparation"]
  node_encode["ENCODE inputs<br/>[encode.py]"]
  node_windows["Sequence windows<br/>[windows.py]"]
  node_negatives["Negative sets<br/>[negatives.py]"]
  node_splits["Fold construction<br/>[splits.py]"]
  node_panel["Study panel<br/>[panel.py]"]
  node_config["Configuration<br/>[config.py]"]
end

subgraph group_models["Modeling"]
  node_train_data["Training data<br/>[data.py]"]
  node_registry["Model registry<br/>[registry.py]"]
  node_cnn["CNN model<br/>[cnn.py]"]
  node_lm["Language models<br/>[lm.py]"]
  node_trainer["Model trainer<br/>[trainer.py]"]
end

subgraph group_analysis["Evaluation and evidence"]
  node_nested["Nested estimands<br/>[nested.py]"]
  node_stats["Statistical methods<br/>[stats.py]"]
  node_delong["AUROC comparison<br/>[delong.py]"]
  node_cloud_analysis["Cloud analysis<br/>[cloud_analysis.py]"]
  node_verify["Evidence verification<br/>[verify.py]"]
end

subgraph group_cloud["Cloud execution"]
  node_ingest["Cloud ingestion<br/>[cloud_ingest.py]"]
  node_cloud_store["Cloud storage access<br/>[cloud.py]"]
  node_local_store[("Local artifact store<br/>[localstore.py]")]
  node_modal_sweep["GPU sweeps<br/>[modal_sweep.py]"]
end

subgraph group_variants["Variant study"]
  node_variant_assign["Variant assignment<br/>[assign.py]"]
  node_conservation["Conservation control<br/>[conservation.py]"]
end

node_researcher(("Researcher"))
node_external_data(("ENCODE and references"))

node_researcher -->|"starts ingestion"| node_ingest
node_external_data -->|"provides inputs"| node_ingest
node_ingest -->|"uploads data"| node_cloud_store
node_encode -.->|"supplies peaks"| node_windows
node_negatives -->|"uses windows"| node_windows
node_splits -->|"uses windows"| node_windows
node_researcher -->|"launches sweeps"| node_modal_sweep
node_modal_sweep -.->|"runs training"| node_trainer
node_train_data -->|"loads folds"| node_splits
node_registry -->|"reads settings"| node_config
node_registry -->|"builds CNN"| node_cnn
node_registry -->|"builds language model"| node_lm
node_trainer -.->|"uses model handle"| node_registry
node_trainer -.->|"consumes loaders"| node_train_data
node_cloud_analysis -.->|"computes estimands"| node_nested
node_nested -->|"uses methods"| node_stats
node_nested -->|"uses AUROC tests"| node_delong
node_cloud_analysis -->|"accesses cloud"| node_cloud_store
node_cloud_analysis -->|"uses variant analysis"| node_conservation
node_verify -->|"checks evidence"| node_cloud_store
node_researcher -->|"checks results"| node_verify
node_variant_assign -->|"uses sequence windows"| node_windows
node_conservation -->|"uses estimators"| node_stats
node_cloud_analysis -->|"returns scores locally"| node_local_store
node_panel -.->|"selects datasets"| node_train_data

click node_encode "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/data/encode.py"
click node_windows "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/data/windows.py"
click node_negatives "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/data/negatives.py"
click node_splits "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/data/splits.py"
click node_panel "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/utils/panel.py"
click node_config "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/utils/config.py"
click node_train_data "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/train/data.py"
click node_registry "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/models/registry.py"
click node_cnn "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/models/cnn.py"
click node_lm "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/models/lm.py"
click node_trainer "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/train/trainer.py"
click node_nested "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/eval/nested.py"
click node_stats "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/stats.py"
click node_delong "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/eval/delong.py"
click node_cloud_analysis "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/scripts/cloud_analysis.py"
click node_verify "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/scripts/verify.py"
click node_ingest "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/scripts/cloud_ingest.py"
click node_cloud_store "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/utils/cloud.py"
click node_local_store "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/utils/localstore.py"
click node_modal_sweep "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/cloud/modal/modal_sweep.py"
click node_variant_assign "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/variants/assign.py"
click node_conservation "https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/variants/conservation.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_encode,node_windows,node_negatives,node_splits,node_panel,node_config toneBlue
class node_train_data,node_registry,node_cnn,node_lm,node_trainer toneAmber
class node_nested,node_stats,node_delong,node_cloud_analysis,node_verify toneMint
class node_ingest,node_cloud_store,node_local_store,node_modal_sweep toneRose
class node_variant_assign,node_conservation,node_researcher,node_external_data toneIndigo
```

## Components

### Data preparation

- [ENCODE inputs](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/data/encode.py): `src/rbp/data/encode.py`
- [Sequence windows](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/data/windows.py): `src/rbp/data/windows.py`
- [Negative sets](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/data/negatives.py): `src/rbp/data/negatives.py`
- [Fold construction](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/data/splits.py): `src/rbp/data/splits.py`
- [Study panel](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/utils/panel.py): `src/rbp/utils/panel.py`
- [Configuration](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/utils/config.py): `src/rbp/utils/config.py`

### Modeling

- [Training data](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/train/data.py): `src/rbp/train/data.py`
- [Model registry](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/models/registry.py): `src/rbp/models/registry.py`
- [CNN model](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/models/cnn.py): `src/rbp/models/cnn.py`
- [Language models](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/models/lm.py): `src/rbp/models/lm.py`
- [Model trainer](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/train/trainer.py): `src/rbp/train/trainer.py`

### Evaluation and evidence

- [Nested estimands](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/eval/nested.py): `src/rbp/eval/nested.py`
- [Statistical methods](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/stats.py): `src/rbp/stats.py`
- [AUROC comparison](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/eval/delong.py): `src/rbp/eval/delong.py`
- [Cloud analysis](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/scripts/cloud_analysis.py): `scripts/cloud_analysis.py`
- [Evidence verification](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/scripts/verify.py): `scripts/verify.py`

### Cloud execution

- [Cloud ingestion](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/scripts/cloud_ingest.py): `scripts/cloud_ingest.py`
- [Cloud storage access](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/utils/cloud.py): `src/rbp/utils/cloud.py`
- [Local artifact store](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/utils/localstore.py): `src/rbp/utils/localstore.py`
- [GPU sweeps](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/cloud/modal/modal_sweep.py): `cloud/modal/modal_sweep.py`

### Variant study

- [Variant assignment](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/variants/assign.py): `src/rbp/variants/assign.py`
- [Conservation control](https://github.com/nirmalkumar31/rbp-protocol-calibration/blob/main/src/rbp/variants/conservation.py): `src/rbp/variants/conservation.py`

### Other components

- **Researcher**
- **ENCODE and references**

## Connections

- Researcher → Cloud ingestion: starts ingestion (evidence: `README.md`)
- ENCODE and references → Cloud ingestion: provides inputs (evidence: `scripts/cloud_ingest.py`)
- Cloud ingestion → Cloud storage access: uploads data (evidence: `scripts/cloud_ingest.py`)
- ENCODE inputs → Sequence windows: supplies peaks
- Negative sets → Sequence windows: uses windows (evidence: `src/rbp/data/negatives.py`)
- Fold construction → Sequence windows: uses windows (evidence: `src/rbp/data/splits.py`)
- Researcher → GPU sweeps: launches sweeps (evidence: `README.md`)
- GPU sweeps → Model trainer: runs training
- Training data → Fold construction: loads folds (evidence: `src/rbp/train/data.py`)
- Model registry → Configuration: reads settings (evidence: `src/rbp/models/registry.py`)
- Model registry → CNN model: builds CNN (evidence: `src/rbp/models/registry.py`)
- Model registry → Language models: builds language model (evidence: `src/rbp/models/registry.py`)
- Model trainer → Model registry: uses model handle
- Model trainer → Training data: consumes loaders
- Cloud analysis → Nested estimands: computes estimands
- Nested estimands → Statistical methods: uses methods (evidence: `src/rbp/eval/nested.py`)
- Nested estimands → AUROC comparison: uses AUROC tests (evidence: `src/rbp/eval/nested.py`)
- Cloud analysis → Cloud storage access: accesses cloud (evidence: `scripts/cloud_analysis.py`)
- Cloud analysis → Conservation control: uses variant analysis (evidence: `scripts/cloud_analysis.py`)
- Evidence verification → Cloud storage access: checks evidence (evidence: `scripts/verify.py`)
- Researcher → Evidence verification: checks results (evidence: `README.md`)
- Variant assignment → Sequence windows: uses sequence windows (evidence: `src/rbp/variants/assign.py`)
- Conservation control → Statistical methods: uses estimators (evidence: `src/rbp/variants/conservation.py`)
- Cloud analysis → Local artifact store: returns scores locally (evidence: `README.md`)
- Study panel → Training data: selects datasets
