InteRD

InteRD infers cell-type proportions from bulk RNA sequencing (RNA-seq) data by integrating single-cell RNA sequencing (scRNA-seq) references to produce robust deconvolution across disease conditions.

Key Features:

• Integration of Multiple scRNA-seq Datasets: Integrates deconvolution results from multiple scRNA-seq datasets to enhance robustness and accuracy of cell-type proportion estimates.
• Calibration with Biological Priors: Calibrates estimates using external biological information as priors, including selected scRNA-seq references.
• Robustness to Inaccurate Information: Maintains robustness against inaccuracies in external reference information to yield reliable estimates with imperfect references.
• Penalized Regression and Evaluation Criterion: Uses penalized regression coupled with a novel evaluation criterion to improve deconvolution performance.
• Implementation: Implemented in R.

Scientific Applications:

• Analysis of disease-relevant tissues: Provides accurate, biologically consistent estimates of cell-type proportions to study variations in cell-type composition across disease conditions.
• Investigation of cellular mechanisms: Facilitates exploration of cell-type dynamics underlying various diseases using integrated scRNA-seq-informed deconvolution.

Methodology:

Integrates deconvolution results from multiple scRNA-seq datasets, calibrates with biological priors (e.g., selected scRNA-seq references), applies penalized regression, uses a novel evaluation criterion, and incorporates measures to remain robust to inaccurate external information.