PKbioanalysis

PKbioanalysis is a comprehensive R package designed to streamline pharmacokinetic (PK) and bioanalytical workflows from study design through data analysis and reporting. Built on regulatory best practices and FAIR principles, it provides an integrated solution for managing bioanalytical experiments with persistent data storage, interactive visualizations, and AI-assisted quality control.

✨ Key Features

📊 Study Management & Design

• Comprehensive trial management system with relational database architecture (DuckDB)
• Study design tools for common PK studies such as single-dose (SD), multiple-dose (MD), food-effect (FE), and bioequivalence (BE) studies along with In Vitro studies support
• Subject tracking with dosing schedules, sampling timepoints, and metadata management
• Sample log integration linking bioanalytical data to study design

🧪 Bioanalytical Workflows

• 96-well plate design and visualization with flexible filling schemes (horizontal/vertical)
• Automated injection sequence generation compatible with major LC-MS platforms
• Vendor support: MassLynx, MassHunter, Analyst
• Interactive chromatogram integration with manual and automated peak detection
• Quality control (QC) Assessment using regulatory-compliant criteria
• Suitability assessment for instrument equilibration monitoring
• Linearity evaluation with interactive visualization and regulatory-compliant reporting

📈 Data Analysis & Export

• Maximum likelihood estimation (MLE) of additive and proportional errors
• Interactive dilution scheme with automatic unit conversion
• PKmerge functionality to combine bioanalytical results with study metadata
• NONMEM-ready export with numeric recoding and codebook generation
• Precision and accuracy calculations per analytical batch

🤖 AI Capabilities

• AI-assisted chromatogram integration with automated peak boundary detection
• Intelligent quality assessment for linearity, suitability, and study design
• Conversational AI assistant for method troubleshooting and data interpretation
• Regulatory compliance checks with automated flagging of potential issues

📦 Installation

GUI-Only Installation (No Coding Required)

PKbioanalysis provides modular Shiny applications for study management (study_app()), chromatography processing (chrom_app()), and quantification (quant_app()). These run locally with persistent data storage.

Windows Users

1. Download the installer and shortcuts from Google Drive
2. Run install_PKbioanalysis.bat to install the package
3. Use the desktop shortcuts:
   • study_app.bat - Study design and sample management
   • chrom_app.bat - Chromatogram integration
   • quant_app.bat - Quantification and linearity

R Package Installation

For users comfortable with R programming:

Stable Release (CRAN)

install.packages("PKbioanalysis")

Development Version (GitHub)

# Install remotes if needed
install.packages("remotes")

# Install PKbioanalysis from GitHub
remotes::install_github("OmarAshkar/PKbioanalysis")

Optional: Python Dependencies

For advanced chromatography file parsing (Waters .raw files):

PKbioanalysis::install_py_dep()

This creates a virtual environment with required Python packages (pandas, rainbow-api, numpy, scipy).

🚀 Quick Start

library(PKbioanalysis)

# Study design and management
study_app()

# Chromatogram integration
chrom_app()

# Quantification, linearity assessment, residual error estimation, and PK dataset generation
quant_app()

🤖 AI Capabilities & Configuration

PKbioanalysis integrates AI-powered quality assessment and decision support throughout the bioanalytical workflow.

Supported AI Features

1. Automated Chromatogram Integration

• AI analyzes chromatographic traces to detect peak boundaries
• Identifies retention time, peak start/end, and signal-to-noise ratio
• Flags problematic peaks with detailed comments
• Validates peak shape and width according to analytical standards

2. Linearity Assessment Assistant

• Reviews calibration curve statistics
• Identifies outliers and recommends exclusions
• Checks intercept significance and heteroscedasticity (recommends weighting if needed)
• Provides regulatory compliance feedback

3. Suitability Evaluation

• Analyzes instrument response stabilization across runs
• Calculates equilibration time based on CV% trends
• Flags experimental issues (insufficient replicates, high variability)

4. Study Design Review

• Evaluates randomization, blocking, and control groups
• Suggests improvements for sampling strategy
• Assesses balance and potential confounding factors in the design

5. Plate Design Optimization

• Reviews QC distribution and calibration curve coverage
• Checks for appropriate controls (blanks, suitability samples)
• Validates replicate strategy

6. Injection List Quality Control

• Analyzes run order and blank placement
• Identifies potential carryover risks
• Suggests optimization for batch structure

AI Configuration

PKbioanalysis uses OpenAI-compatible APIs (including local models via Ollama or cloud providers).

Setup via GUI

1. Launch any app (study_app(), chrom_app(), or quant_app())
2. Click the ⚙️ Configure Settings button
3. Enter your configuration:
   • API Base URL: https://api.openai.com/v1 or your local endpoint
   • API Key: Your OpenAI API key (or leave blank for local models)
   • AI Model: Choose from supported models
   • Temperature: Control response randomness (0.0 = deterministic, 1.0 = creative)

Setup Programmatically

# Update configuration
PKbioanalysis::update_config(
  base_url = "https://api.openai.com/v1",
  api_key = Sys.getenv("OPENAI_API_KEY"),  # Or set in .Renviron
  model = "gpt-4",
  temperature = 0.5
)

# Refresh to apply changes
PKbioanalysis::refresh_config()

# Check current settings
PKbioanalysis::get_pkbioanalysis_option("ai_model")

Supported Models

The package supports any OpenAI-compatible model, including: - OpenAI: gpt-4, gpt-3.5-turbo - Open-source via Ollama/LM Studio: llama-3.1-70b-instruct, mistral-7b-instruct, codestral-22b - Cloud providers: gemma-3-27b-it, granite-3.3-8b-instruct

Environment Variables (Alternative Setup)

# In .Renviron file
OPENAI_API_KEY=your_api_key_here

Using Local Models (Privacy-First)

For organizations requiring data privacy: 1. Install Ollama or LM Studio 2. Download a model (e.g., ollama pull llama3.1:70b) 3. Configure PKbioanalysis:

update_config(
  base_url = "http://localhost:11434/v1",  # Ollama default
  api_key = "not-needed",  # Local models don't need keys
  model = "llama3.1:70b"
)

AI Usage Tips

• Higher temperature (0.7-1.0) for creative suggestions and exploratory analysis
• Lower temperature (0.0-0.3) for consistent, deterministic quality checks
• Larger models (70B parameters) for complex regulatory assessments
• Smaller models (7-8B parameters) for routine peak integration and QC

Documentation & Resources

• Package Website: https://omarashkar.github.io/PKbioanalysis/

Architecture

PKbioanalysis uses a relational database (DuckDB) to maintain study integrity:

Study Design → Plate Design → Injection Sequences → Chromatography → Quantification
     ↓              ↓                 ↓                   ↓              ↓
  Subjects      Samples           File List           Peak Data    Concentrations
     ↓              ↓                 ↓                   ↓              ↓
  Dosing         Metadata          Database           Linearity      PK Datasets

License

AGPL-3.0 or later. See details.