You are currently browsing the tag archive for the ‘Batch effects’ tag.

A storm is brewing in the last week or so over the results in a study published in Science on genes related to longevity. At the heart of this are key issues in experimental design and quality control. A story of two platforms, genotyping calling errors and at least two variants that appear to be false positives – SNP Gate. However, what I’m most curious about is how many scientists will read the media snapshots of this and actually see some points that can help them in their own study:

  1. Batch effects can introduce noise into a study, possibly confounding interpretation of the results. To avoid this, possible issues need to be taken into consideration BEFORE the study starts. In this case, it might have been wise to order/run all the arrays at one time, rather than to start running arrays and find out the manufacturer stopped making them before you complete your study.
  2. Studies can live and die by the genotyping. QA/QC, examination of impact of genotyping algorithm – all of these are steps that need to happen before the analysis begins.
  3. Replication is still King (sorry LeBron). No matter how persuasive the results look, hold your breath until you have replication in an independent population.

    Photo by Roberto Hurtado