IBRAP

IBRAP performs integrated benchmarking and analysis of single-cell RNA-sequencing (scRNA-seq) datasets to evaluate and compare preprocessing, quality control, normalization, dimensionality reduction, integration, and clustering workflows.

Key Features:

• Pre-processing: Implements pre-processing steps for scRNA-seq datasets.
• Quality control: Provides quality-control procedures for single-cell transcriptomic data.
• Normalization: Supports normalization methods for scRNA-seq data.
• Dimensionality reduction: Performs dimensionality reduction on high-dimensional single-cell data.
• Integration: Supports integration across single and multiple samples for combined analyses.
• Clustering: Performs clustering to identify cell populations from scRNA-seq data.
• Interchangeable analytical components: Offers a suite of interchangeable analytical components that can be tailored to specific datasets.
• Integrated benchmarking: Includes benchmarking functionality with multiple metrics to compare pipeline configurations and evaluate performance across diverse data types.
• Annotation modes: Supports both reference-based cell annotation and unsupervised analysis.

Scientific Applications:

• Primary pancreatic tissue integration: Single- and multi-sample integration analyses have been performed on primary pancreatic tissue datasets.
• Cancer cell line analysis: Integration and benchmarking have been applied to cancer cell line scRNA-seq datasets.
• Simulated datasets with ground truth: Benchmarking using simulated datasets with ground-truth cell labels enables direct evaluation of pipeline accuracy.
• Reference map construction: Integration of multiple samples and studies to construct reference maps of normal and diseased tissues.
• Cell-type identification: Comparative evaluations demonstrate the use of reference-based annotation to identify major and minor cell types.

Methodology:

Computational steps explicitly include pre-processing, quality control, normalization, dimensionality reduction, integration, clustering, benchmarking with multiple metrics, and both reference-based and unsupervised cell annotation.