2 vs. 4 haplotype results

[dmartche]

trafficstars

Hello, When running MSMC2 on 2 vs. 4 haplotypes I get extremely high Ne estimates in the 2 haplotype runs. The flags for the runs are -s -I 0,1 (Individual 1) ; -s -I 2,3 (Individual 2) and -s -I 0,1,2,3 (Individuals 1 and 2). What could be causing this? Any feedback is greatly appreciated!

[stschiff]

Very odd... could be phasing, which is ignored for -I 0,1 and -I 2,3 but not for -I 0,1,2,3. But hard to say in general.

[dmartche]

The results for phased and unphased data are almost identical. Could this be an issue with the phasing itself? On a related note, the results for the 4 haplotype runs are more inline with the PSMC results vs. the 2 haplotype runs. This pattern of very high values with 2 haplotype MSMC2 (4-7mln Ne) compared to PSMC (<0.2 mln Ne) is also seen in Robinson et al. 2021 (Fig S3B) and Nieto-Blázquez et al. 2022 (Fig S2).