Research article

Meta analysis of whole-genome linkage scans with data uncertainty: an application to Parkinson's disease

Albert Rosenberger^1* , Manu Sharma^2* , Bertram Müller-Myhsok³ , Thomas Gasser² and Heike Bickeböller¹

¹  Georg-August-University Göttingen, Medical School, Department of Genetic Epidemiology, Germany

²  Eberhard-Karl-University Tübingen, Centre of Neurology, Hertie Institute for Clinical Brain Research, Germany

³  Max-Planck Institute for Psychiatry, Munich, Germany

author email corresponding author email* Contributed equally

BMC Genetics 2007, 8:44doi:10.1186/1471-2156-8-44

Published:	2 July 2007

Abstract

Background

Genome wide linkage scans have often been successful in the identification of genetic regions containing susceptibility genes for a disease. Meta analysis is used to synthesize information and can even deliver evidence for findings missed by original studies. If researchers are not contributing their data, extracting valid information from publications is technically challenging, but worth the effort. We propose an approach to include data extracted from published figures of genome wide linkage scans. The validity of the extraction was examined on the basis of those 25 markers, for which sufficient information was reported. Monte Carlo simulations were used to take into account the uncertainty in marker position and in linkage test statistic. For the final meta analysis we compared the Genome Search Meta Analysis method (GSMA) and the Corrected p-value Meta analysis Method (CPMM). An application to Parkinson's disease is given. Because we had to use secondary data a meta analysis based on original summary values would be desirable.

Results

Data uncertainty by replicated extraction of marker position is shown to be much smaller than 30 cM, a distance up to which a maximum LOD score may usually be found away from the true locus. The main findings are not impaired by data uncertainty.

Conclusion

Applying the proposed method a novel linked region for Parkinson's disease was identified on chromosome 14 (p = 0.036). Comparing the two meta analysis methods we found in this analysis more regions of interest being identified by GSMA, whereas CPMM provides stronger evidence for linkage. For further validation of the extraction method comparisons with raw data would be required.