Do we measure or compute polygenic risk scores? Why language matters

Area of Science:

• Genetics
• Biomarkers
• Statistical modeling

Background:

• Biomarkers such as blood pressure and sodium levels are quantifiable and visualizable.
• Polygenic risk scores (PRSs) are statistical constructs derived from genetic variants.
• The framing and naming of PRSs influence their interpretation and perceived value.

Purpose of the Study:

• To differentiate polygenic risk scores (PRSs) as distinct epistemic objects from traditional biomarkers.
• To analyze the implications of how PRSs are framed and named.
• To explore the tangible versus inferential understanding of PRSs and their associated languages of measurement and computation.

Main Methods:

• Conceptual analysis distinguishing between measurement and computation.
• Examination of the epistemic status of PRSs.
• Critical review of current practices in polygenic risk score 'doing'.

Main Results:

• PRSs are fundamentally different from traditional biomarkers due to their computational and assembled nature.
• Conflating tangible and inferential frames of PRSs obscures critical questions about their representation and validity.
• Current methods for 'doing PRS' may not be optimal or fully responsible.

Conclusions:

• Clarifying the epistemic nature of PRSs is essential for scientific rigor.
• Further investigation is needed into the optimal and responsible methodologies for PRS calculation and application.
• The credibility of PRS-based knowledge claims requires careful consideration of their inferential basis.