A genetic scan includes noise that changes with the environment
A new statistical method searches large biobank datasets for genetic effects that vary with measured environmental conditions. The peer-reviewed fastGxE study models changing noise alongside genetic background, then applies the approach to physical traits and blood biomarkers. Many associations survived an internal split-sample check. The observations concern one ancestry group and require separate population replication before broader interpretation.
Science··Morning
An internal check supports many associations
Researchers divided a biobank dataset into two separate halves to check a new method for finding genetic effects that vary with the environment. After correction for multiple testing, the second half supported 21 physical-trait and 36 blood-biomarker associations. This was an internal check within one resource. The peer-reviewed technical study calls the method fastGxE and combines observed exposures with genetic information in a statistical model.[1]
Changing noise can resemble an interaction
An apparent change in a genetic association can arise when measurement variability changes across exposure levels. The model includes this environment-dependent noise alongside genetic main effects and the contributions of many variants. Applications covered 42 environmental variables, alongside 32 physical traits and 67 blood biomarkers, including sleep, smoking, diet and physical activity. The analysis identified 23 interaction associations across seven genomic regions for physical traits and 38 across 18 regions for blood biomarkers.[1]
The population defines the reach of the result
UK Biobank, a large British health-research resource, supplied a quality-controlled genotype pool of 430,145 participants identifying as white British. Each trait analysis used fewer people where measurements were missing. The exposures were observed, leaving intervention effects outside the association analysis. Extra false-positive signals appeared in some simulations with rare proxy variants and strong main genetic effects. The code is public, while individual participant data require approved access and independent population replication remains outstanding.[1]
Related columns
For more information on this topic, you can read the related columns.
