ClustAll: An R package for patient stratification in complex diseases

Area of Science:

• Computational biology
• Bioinformatics
• Genomics

Background:

• Precision medicine requires understanding patient heterogeneity for tailored treatments.
• Complex diseases present challenges in patient stratification due to data intricacies.

Purpose of the Study:

• To introduce ClustAll, a Bioconductor package for unsupervised patient stratification.
• To provide a robust framework for handling mixed data types, missing values, and collinearity in clinical data.
• To enable identification of multiple, robust patient stratifications.

Main Methods:

• ClustAll utilizes a validated clustering framework adapted for clinical data.
• The package incorporates S4 classes and parallel computing for efficiency.
• User-friendly tools are provided for exploring and comparing stratifications.

Main Results:

• ClustAll effectively handles mixed data types, missing values, and collinearity.
• The package can identify multiple, robust patient stratifications within a population.
• Validation on public clinical datasets confirms its effectiveness.

Conclusions:

• ClustAll is a powerful tool for patient stratification in personalized medicine.
• The package enhances computational efficiency and user experience.
• It has the potential to significantly impact clinical management strategies.