gms | German Medical Science

MAINZ//2011: 56. GMDS-Jahrestagung und 6. DGEpi-Jahrestagung

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V.
Deutsche Gesellschaft für Epidemiologie e. V.

26. - 29.09.2011 in Mainz

Imputation-free GWAS Meta-Analysis with YAMAS

Meeting Abstract

  • Tim Becker - DZNE, Bonn
  • Dmitriy Drichel - DZNE, Bonn
  • Christine Herold - DZNE, Bonn
  • Christian Meesters - DZNE, Bonn
  • Markus Leber - IMBIE, Bonn

Mainz//2011. 56. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie (gmds), 6. Jahrestagung der Deutschen Gesellschaft für Epidemiologie (DGEpi). Mainz, 26.-29.09.2011. Düsseldorf: German Medical Science GMS Publishing House; 2011. Doc11gmds053

doi: 10.3205/11gmds053, urn:nbn:de:0183-11gmds0531

Published: September 20, 2011

© 2011 Becker et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( You are free: to Share – to copy, distribute and transmit the work, provided the original author and source are credited.



The success of genome-wide association studies (GWAS) is motivation enough to exchange and combine available data resources to expediently discover genetic risk factors for common human traits. The primary tools for these efforts are imputation and sub-sequent meta-analysis (MA).

A main purpose of imputation prior to MA is to unify the available marker panels and to avoid loss of SNPs that are present in one study but not in another. With YAMAS – Yet Another Meta-Analysis Software – we present a MA software that avoids such loss without the need to impute data. By using reference data from the HAPMAP and 1000 genomes projects users are enabled to analyze all SNPs that are present in at least one of the experimental marker panels: LD-information is used to find substitute makers for those missing. For each SNP that is missing in one study, the marker from the study with largest r², according to HAPMAP data, with the missing marker is chosen as a “proxy SNP”. Furthermore, based on HAPMAP haplotype frequencies, “proxy alleles” of a SNP and its proxy-SNP are identified, i.e. it is decided which allele of the missing SNP predominantly occurs in combination with which allele of the proxy-SNP. For each SNP present in at least one study, MA of that SNP is done by combining the association results of the SNP or, if not available the proxy-SNP, across studies. Consistency of the direction of effects is automatically accounted for through the proxy allele definition.

We present results from a power simulation study and compare the relative performance of MA with imputation and imputation-free MA with YAMAS. Not surprisingly, MA with our algorithm in general is not quite as powerful as MA with imputation, since the marker panel is still smaller than with imputing. However, in a variety of scenarios, the power loss with the proxy-algorithm is only moderate.

In summary, MA with the YAMAS proxy-algorithm is a quick and easy alternative, yielding ad hoc results and thereby giving an incentive to follow-up analysis. Keeping in mind that collaborative MA efforts are frequently long-lasting, because all participating groups have to provide imputing analysis results in a coherent fashion, the time advantage becomes apparent.