Usage: any benefit of providing a genetic map estimated from the same reference panel?

[Gullumluvl]

Hi!

I would like to use GLIMPSE2 for imputation in a non-model mammal species. I have a panel of high-coverage genomes (about a hundred individuals).

The documentation states that without a genetic map, a constant recombination rate is assumed across the genome.

• My question is, is there any information gained from building a recombination map from the same reference panel? i.e, doesn't GLIMPSE2 already incorporates the same linkage information from the input reference panel, as what would be inferred by a recombination map tool (LDhelmet, pyrho, etc)? These latter tools use additional information, such as a theta estimate, a past demography inference, and simulations of coalescence, but I am not so sure of the increase in accuracy with regard to the overall uncertainty. I am sorry, I do not know the internal working of GLIMPSE2 so I cannot answer this myself.
• A complementary question is: are we not introducing too much confidence in a given recombination landscape by doing a preliminary map inference on the same panel?

Thanks in advance for your help!

Best,