Pipeline for GWAS Using MLMM
Pipeline for Genome-Wide Association Study using Multi-Locus Mixed Model from Segura V, Vilhjálmsson BJ et al. (2012) <10.1038>. The pipeline include detection of associated SNPs with MLMM, model selection by lowest eBIC and p-value threshold, estimation of the effects of the SNPs in the selected model and graphical functions.10.1038>
mlmm.gwas v1.0.2 (2018/03/28)
Initial release on CRAN
mlmm.gwas v1.0.3 (2018/05/03, Olivier Guillaume)
- Correction of a bug in eBIC_allmodels with FM or FMI models.
- Correction of a bug in the manhattan.plot function.
- R3.5 support
mlmm.gwas v1.0.4 (2018/07/18, Olivier Guillaume and Alexandra Duhnen)
Important changes (bug correction and changes of functionality):
- BLUE estimation in Estimation_allmodels have been modified for more coherence, with same constraints for all effects (AD)
- Correction of a bug in the genotypes.boxplot function, the tukey classes symbols were wrong in some cases (OG)
- Correction of a bug in the run_entire_gwas_pipeline function, cofactors were not used for effects estimation (AD)
- The package dependency is now R (>= 3.3.0) instead of R (>= 3.4.0).
- Chromosome are now ordered in alphanumerical order in Manhattan plot instead of alphabetical order (ie. 10 now comes after 9). (OG)
- Special characters in marker names are now supported, especialy ":" that use to cause problem. (OG)
- Unwanted graphical windows opened during mlmm_allmodels execution are now closed immediatly. The window is opened by the sommer:mmer2 function. A way to remove it entierely is yet to be found. (OG)
- The marker order in the table generated by the Estimation_allmodels is now the same as the marker selection order in MLMM instead of alphabetical order (OG)
- Correction of a small bug in the Estimation_allmodels function were underscore were injected in cofactor names (OG)