# victor369basu/proteinstructureprediction architecture

> Architecture diagram of the GitHub repository victor369basu/proteinstructureprediction, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2025-01-27). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/victor369basu/proteinstructureprediction
- Repository: https://github.com/victor369basu/proteinstructureprediction

## Overview

Key Components to Include in the System Design:

##### Data Flow Components
- Input Processing Layer
- Sequence Input Path (for sequence-only model)
- PSSM Input Path (for PSSM-based model)
- Model Core
- Output Processing Layer
- Visualization Pipeline

##### Main Architectural Components

a. Input Processing Module:
- Path 1: Protein Sequence Input (dimension: ℝ¹)
- Path 2: PSSM + Secondary Structure + Information Content (dimension: ℝ⁴⁹)

- Attention Layer (from attention_layer.py)
- Embedding Layer
- Multi-Head Attention Block
- Hidden Layers

c. Output Processing:
- Angle Prediction Layer (12 angles × 2 for sin/cos)
- Coordinate Conversion Module

d. Training Pipeline:
- Loss Calculation
- Optimization Loop

e. Visualization System:
- 3D Structure Rendering
- PDB File Generation

Design Guidelines:

- Double lines: Training flow

##### Technical Details
- Input dimensions (49 for PSSM, 1 for sequence)
- Number of attention heads (8)
- Hidden dimensions (512)
- Embedding dimension (32)

**[Input Layer] → [Embedding] → [Attention Block] → [Output Layer] → [Visualization]**
- ↑              ↑              ↑                    ↓              ↓
**[Config] -----> [Training Loop] <---- [Loss Calculation]   [3D Rendering]**

## Diagram (Mermaid)

```mermaid
graph TB
    %% Styles and Classes
    classDef input fill:#a8d1f0,stroke:#000
    classDef model fill:#90EE90,stroke:#000
    classDef output fill:#FFE5B4,stroke:#000
    classDef viz fill:#DDA0DD,stroke:#000
    classDef config fill:#D3D3D3,stroke:#000
    classDef training fill:#FFA07A,stroke:#000

    %% Input Processing Layer
    subgraph InputLayer
        SeqInput["Sequence Input<br/>(ℝ¹)"]:::input
        PSSMInput["PSSM + Secondary Structure<br/>+ Info Content (ℝ⁴⁹)"]:::input
    end

    %% Model Core
    subgraph ModelCore
        Embed["Embedding Layer<br/>(dim: 32)"]:::model
        Attention["Attention Layer<br/>(8 heads)"]:::model
        Hidden["Hidden Layers<br/>(dim: 512)"]:::model
    end

    %% Output Processing
    subgraph OutputLayer
        AnglePred["Angle Prediction<br/>(12 angles × 2)"]:::output
        CoordConv["Coordinate<br/>Conversion"]:::output
    end

    %% Training Components
    subgraph TrainingSystem
        Loss["Loss<br/>Calculation"]:::training
        Optimizer["Optimization<br/>Loop"]:::training
    end

    %% Visualization
    subgraph VisualizationSystem
        Render["3D Structure<br/>Rendering"]:::viz
        PDBGen["PDB File<br/>Generation"]:::viz
    end

    %% Configuration
    Config["Configuration<br/>Parameters"]:::config

    %% Connections
    SeqInput --> Embed
    PSSMInput --> Embed
    Embed --> Attention
    Attention --> Hidden
    Hidden --> AnglePred
    AnglePred --> CoordConv
    CoordConv --> Render
    Render --> PDBGen
    
    AnglePred --> Loss
    Loss --> Optimizer
    Optimizer --> |Update| Attention
    Config --> |Parameters| ModelCore
    Config --> |Settings| TrainingSystem

    %% Click Events
    click Attention "https://github.com/victor369basu/ProteinStructurePrediction/blob/master/attention_layer.py"
    click ModelCore "https://github.com/victor369basu/ProteinStructurePrediction/blob/master/model.py"
    click TrainingSystem "https://github.com/victor369basu/ProteinStructurePrediction/blob/master/trainer.py"
    click VisualizationSystem "https://github.com/victor369basu/ProteinStructurePrediction/blob/master/visualize.py"
    click Config "https://github.com/victor369basu/ProteinStructurePrediction/blob/master/config.py"
    click CoordConv "https://github.com/victor369basu/ProteinStructurePrediction/blob/master/helper.py"
    
    %% Legend
    subgraph Legend
        L1["Input Components"]:::input
        L2["Model Components"]:::model
        L3["Output Components"]:::output
        L4["Visualization"]:::viz
        L5["Training"]:::training
        L6["Configuration"]:::config
    end
```
