Some benchmark papers might be a good starting point to this investigation:
- Accuracy with rare variants, featuring their own work
GIGI and Ped_Pop : https://genome.cshlp.org/content/29/1/125.full.pdf notice here Beagle is version 4.1. In fact here it shows that merlin does a decent job with dense markers. Notice that phasing is performed by MORGAN
- Benchmark on cattle pedigree, featuring Beagle 5.2 https://bmcgenomics.biomedcentral.com/articles/10.1186/s12864-022-08354-6
Also this paper seems promising but it did not compare with other family based methods such as MERLIN or GIGI.
At this point, it is unclear what to use. MORGAN, like MERLIN, is based on Lander-Green for small pedigrees but a hybrid MCMC based on both Elston-Stewart and Lander-Green to handle large pedigrees. However, Lander-Green works under the linkage equilibrium assumption (which does not hold for dense markers) and unlike MERLIN, it is not clear how (or if!) MORGAN/GIGI handle LD. I need to read more into these.
Some benchmark papers might be a good starting point to this investigation:
GIGIandPed_Pop: https://genome.cshlp.org/content/29/1/125.full.pdf notice here Beagle is version 4.1. In fact here it shows that merlin does a decent job with dense markers. Notice that phasing is performed by MORGANAlso this paper seems promising but it did not compare with other family based methods such as MERLIN or GIGI.
At this point, it is unclear what to use. MORGAN, like MERLIN, is based on Lander-Green for small pedigrees but a hybrid MCMC based on both Elston-Stewart and Lander-Green to handle large pedigrees. However, Lander-Green works under the linkage equilibrium assumption (which does not hold for dense markers) and unlike MERLIN, it is not clear how (or if!) MORGAN/GIGI handle LD. I need to read more into these.