Skip to content

on_boarding.md

Latest commit

 

History

History
69 lines (43 loc) · 4.57 KB

on_boarding.md

File metadata and controls

69 lines (43 loc) · 4.57 KB
graph LR
    User_Interface_CLI_["User Interface (CLI)"]
    Core_Data_Management["Core Data Management"]
    Data_Preparation_for_ML["Data Preparation for ML"]
    Analysis_Visualization["Analysis & Visualization"]
    User_Interface_CLI_ -- "Initiates operations in" --> Core_Data_Management
    User_Interface_CLI_ -- "Initiates operations in" --> Data_Preparation_for_ML
    Core_Data_Management -- "Provides processed data to" --> Data_Preparation_for_ML
    Core_Data_Management -- "Provides data to" --> Analysis_Visualization
    click User_Interface_CLI_ href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/ProteinFlow/User_Interface_CLI_.md" "Details"
    click Core_Data_Management href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/ProteinFlow/Core_Data_Management.md" "Details"
    click Data_Preparation_for_ML href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/ProteinFlow/Data_Preparation_for_ML.md" "Details"
    click Analysis_Visualization href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/ProteinFlow/Analysis_Visualization.md" "Details"
Loading

CodeBoardingDemoContact

Details

The ProteinFlow project is structured around a streamlined pipeline for acquiring, processing, organizing, and preparing protein data for machine learning tasks, complemented by analysis and visualization capabilities. The architecture is designed to facilitate efficient handling of large biological datasets.

User Interface (CLI) [Expand]

The primary command-line interface that serves as the entry point for users to initiate and control the entire data pipeline. It orchestrates the execution of data acquisition, processing, and organization workflows.

Related Classes/Methods:

Core Data Management [Expand]

This foundational component is responsible for the acquisition of raw protein data (PDB and SAbDab files), defining the core data structures for representing proteins and associated ligands, and performing the initial processing steps. This includes filtering, cleaning, and converting raw data into standardized ProteinEntry objects, handling quality checks, and managing ligand-specific details.

Related Classes/Methods:

  • proteinflow.data (1:1)
  • proteinflow.data.PDBEntry (1:1)
  • proteinflow.data.SAbDabEntry (1:1)
  • proteinflow.download (1:1)
  • proteinflow.processing (1:1)
  • proteinflow.ligand (1:1)

Data Preparation for ML [Expand]

Focuses on organizing and partitioning the processed protein data into distinct train, validation, and test sets, often employing clustering techniques to ensure diverse and representative splits. It also provides PyTorch-compatible Dataset and DataLoader classes for efficient batching and preparation of data, making it ready for machine learning model training and evaluation.

Related Classes/Methods:

Analysis & Visualization [Expand]

Offers a comprehensive suite of tools for calculating various protein-related metrics (e.g., sequence similarity, language model perplexity) and for visualizing protein structures and animations. This component also manages optional external dependencies required for its advanced functionalities.

Related Classes/Methods:

FAQ