Statistical method for modeling sequencing data from different technologies in longitudinal studies with application...

Area of Science:

• Genomics
• Bioinformatics
• Biostatistics

Background:

• Longitudinal studies rely on gene expression measurements to track disease severity.
• Technical variations in measurement technology between time points can introduce confounding factors.
• Huntington disease research faces challenges in replicating gene expression findings due to technology shifts.

Purpose of the Study:

• To model the relationship between DeepSAGE and RNA-Seq gene expression technologies.
• To assess the replicability of gene expression-disease severity associations across different measurement platforms.
• To identify sources of statistical inefficiency in longitudinal gene expression studies with technology changes.

Main Methods:

• Utilized gene expression data from Huntington disease patients at two time points.
• Employed DeepSAGE and RNA-Seq technologies, with RNA-Seq used at both time points.
• Applied linear mixed models to jointly analyze gene expression and disease severity, accounting for technology differences.

Main Results:

• Only one of 14 initially significant genes showed consistent association with disease severity across both technologies and time points.
• Significant disagreements were observed between DeepSAGE and RNA-Seq measurements.
• Statistical efficiency was reduced due to inter-technology discordance and measurement error.

Conclusions:

• Replication of gene expression findings in longitudinal studies is challenging when measurement technologies change.
• Joint modeling can elucidate technology-specific effects and their impact on disease association studies.
• Careful consideration of technological differences is crucial for accurate interpretation of longitudinal gene expression data.