Multispectral Data Analysis and Visualization.

PhenoSpectra

PhenoSpectra is an R package designed for processing, analyzing, and visualizing spectral data collected from 3D laser-based scanning systems. This package supports data from various domains, including agriculture, forestry, environmental monitoring, industrial quality control, and biomedical research.

Key Features

• Data Handling:

  • Automated reading, merging, and preprocessing of spectral datasets.
  • Splitting timestamp columns into separate date and time components.
  • Group-based assignment of biological replicates (Bio.Rep).

• Quality Control:

  • Detection of missing values and outliers.
  • Flexible handling of irregularities with options to replace, remove, or impute data.
  • Group-specific outlier detection using statistical methods.

• Feature Selection & Prediction:

  • Selection of the most discriminative spectral variables using FDR correction.
  • Prediction of Spectral Disease Severity (SDS) using linear regression models.

• Customizable Workflow:

  • Supports group-based analyses for treatments or other categorical variables.
  • Generates clean, publication-ready datasets and summary reports.

Installation

To install the PhenoSpectra package directly from GitHub:

1. Ensure you have the devtools package installed:

   install.packages("devtools")
   

2. Install PhenoSpectra:

   devtools::install_github("bayer-int/PhenoSpectra")
   

File Structure

Below is an overview of the directory structure of the PhenoSpectra package:

• R: R functions (e.g., reads.R, qaqcs.R, feature_selection.R, predict_SDS.R)
• man: Documentation files (auto-generated by Roxygen2)
• data: Example datasets (if applicable)
• Demo: Demo input/output files for testing functions
• DESCRIPTION: Package metadata
• NAMESPACE: Package imports and exports
• README.md: Project overview and usage
• LICENSE: Licensing information
• inst: Installation-related files (if required)
• vignettes: Extended examples and tutorials (if applicable)

Usage Examples

1. Using reads()

merged_data <- reads(
  directory = "Demo",
  pattern = "input",
  output_path = "Demo/processed_data.xlsx"
)

2. Using qaqcs()

result <- qaqcs(
  file_path = "Demo/raw_data.xlsx",
  output_path = "Demo/cleaned_data.xlsx",
  handle_missing = "impute",
  handle_outliers = "impute",
  group_by_col = "treatment"
)

# Access the cleaned data and summary table
cleaned_data <- result$cleaned_data
summary_table <- result$summary_table

3. Using feature_selection()

selected_features <- feature_selection(
  file_path = "Demo/cleaned_data.xlsx",
  output_path = "Demo/selected_features.xlsx",
  fdr_threshold = 0.01
)

# View the selected features
print(selected_features)

4. Using predict_SDS()

predicted_sds <- predict_SDS(
  cleaned_data = cleaned_data,
  sf_test = selected_features,
  fixed_effects = c("Scan.date")
)

# View predictions
print(predicted_sds)

License

This project is licensed under the MIT License. See the LICENSE file for more details.

Contact

Dr. Medhat Mahmoud,
Statistical Scientist
Decision Pipeline & Analytics