25 May 2017 09:30am to 10:30am

Meta-analysis with a general genetic model: ACTN3 & athletic performance

Seminar
Event Location
Australia
Speakers
Dr Damjan Vukcevic
Damjan is a Senior Lecturer in Statistical Genomics at the University of Melbourne, an Honorary Fellow at the Murdoch Children’s Research Institute, and the President of the Victorian Branch of...

The ACTN3 gene, known as ‘the gene for speed’, encodes a protein expressed in fast-twitch muscle fibres. About 20% of the population do not express the protein due to a loss-of-function mutation. Presence of one or two copies of this mutation is associated with reduced strength, power and speed, and is underrepresented amongst elite sprint/power athletes. While at least 15 studies have documented this underrepresentation, no consistent model has been used to analyse the data and current meta-analyses are limited by assumptions about the underlying genetic model.

We analyse the data from these studies to assess the three possibilities for ACTN3: either two copies of functional variant (RR), two copies of the loss-of-function variant (XX), or one of each (RX). Rather than compare RR with RX/XX (a dominant model), or RR/RX with XX (a recessive model), each of which entails a strong assumption that might be inconsistent with the data, we use a general model analogous to a two-dimensional version of random-effects meta-analysis. This allows for a straightforward analysis without requiring multiple-testing (with different simpler models). We use proper prior distributions to cope with low counts, which previously led to unnecessary model simplification.

Applying our method to the elite athlete data, we confirm that there is clear evidence that sprint/power athletes have a higher proportion of the functional variant compared to controls, but also show there is substantial heterogeneity in the effect of RX. Neither a recessive nor a dominant model provide an adequate fit to the data, showing the necessity of our more general model. We also show how assuming a misspecified simpler model induces spurious heterogeneity, as exhibited by previous meta-analyses of this variant.