SimBu

SimBu simulates pseudo-bulk RNA-seq datasets by aggregating single-cell RNA-seq expression profiles to produce in silico gold standards for evaluating deconvolution methods and modeling cell-type-specific mRNA bias.

Key Features:

• Profile aggregation: Aggregates single-cell RNA-seq expression profiles in predefined proportions to generate pseudo-bulk RNA-seq datasets with controlled cell-type fractions.
• mRNA bias modeling: Models cell-type-specific mRNA bias using experimentally derived or data-driven scaling factors.
• Controlled cell-type fractions: Provides precise control over simulated cell-type proportions for benchmarking.
• Customizable scenarios: Supports simulation of varying experimental conditions and requirements to test specific aspects of deconvolution algorithms.
• In silico gold standards: Produces datasets intended for benchmarking and systematic evaluation of deconvolution methods.
• Implementation: Provided as an R package.

Scientific Applications:

• Benchmarking deconvolution methods: Evaluates performance of computational deconvolution tools using datasets with known cell-type compositions.
• Assessing mRNA content effects: Studies the impact of cell-type-specific mRNA bias on deconvolution accuracy and on methods for estimating mRNA content.
• Testing algorithm robustness: Simulates tailored scenarios to probe specific features and limitations of deconvolution algorithms.
• Alternative to experimental gold standards: Generates scalable in silico gold standards when techniques such as flow cytometry or immunohistochemistry are impractical.

Methodology:

SimBu aggregates single-cell RNA-seq expression profiles in predefined proportions and applies experimentally derived or data-driven scaling factors to model cell-type-specific mRNA bias; the software is implemented as an R package.