When a research group wants to identify genomic variants associated with a disease or some other trait, the standard approach is to perform a genome-wide association study (GWAS). In this approach, thousands or tens of thousands of individuals are genotyped using genome-wide SNP arrays. After that, the data are imputed to cover even more genetic variation and then statistically analysed. If successful, genomic regions showing association with the trait in question are identified. These associated regions can, however, contain hundreds or even thousands of highly correlated SNPs and the question remains: Which of the associated SNPs have a direct effect and are thus biologically important?
The FINEMAP software is developed to make the statistical process of fine-mapping, i.e. pinpointing the variants that have a direct effect, more computationally efficient. Instead of using exhaustive search like the previously available methods, FINEMAP implements a stochastic search algorithm. Another important feature of the software is that the original genotype data are not needed, since FINEMAP utilises GWAS summary statistics.
We were able to show that FINEMAP produces as accurate results as the more computationally exhaustive programs in a fraction of processing time of these approaches, says Christian Benner, a PhD student at FIMM, who has developed the software.
These features make the software very valuable to researches analysing ever increasing amount of genetic variation data, continues Matti Pirinen, an Academy Research Fellow at FIMM who has supervised the study.
We believe that FINEMAP can reveal valEfficient variable selection using summary data from genome-wide association studiesuable information that could otherwise remain undetected due to computational limitations.
FINEMAP is freely available at: http://www.christianbenner.com/
Original publication:
Benner C, Spencer CC, Havulinna AS, Salomaa V, Ripatti S, Pirinen M. FINEMAP: efficient variable selection usingsummary data from genome-wide associationstudies. Bioinformatics, 32(10), 2016, 1493–1501. doi: 10.1093/bioinformatics/btw018