FAVSeq

FAVSeq identifies technical factors that cause variability between matched bulk and single-cell RNA sequencing (RNA-Seq) measurements by applying machine learning to detect assay-dependent discrepancies in gene expression.

Key Features:

• Systematic Assessment: Evaluates matched single-cell and bulk RNA-Seq datasets subjected to identical pre-processing and sample preparation protocols.
• Machine Learning Integration: Applies machine learning algorithms to discern patterns and factors influencing gene expression variability between technologies.
• Identification of Variability Factors: Identifies contributors to measurement discrepancies including 3’-untranslated region (3’-UTR) length, transcript length, and cellular compartment associations.
• Dropout Analysis: Analyzes dropout events—instances where gene expression is not detected despite being present—particularly prevalent in scRNA-Seq, to distinguish technical dropouts from low expression.

Scientific Applications:

• Distinguishing Technical from Biological Signal: Separates assay-based technical variability from true biological differences in gene expression analyses.
• Refining Expression Analyses: Improves interpretation of gene expression data to reduce false positive discoveries driven by technical artifacts.
• Comparative Transcriptomics: Supports studies comparing single-cell and bulk RNA-Seq to characterize assay-specific biases and their impact on results.

Methodology:

Performs detailed analysis of matched bulk and single-cell datasets and applies machine learning algorithms to detect factors affecting gene expression measurements and dropout occurrences beyond biological differences, low expression levels, or random noise.