Peicheng Jing posted a comment on a wiki page

Dr. Han: I have genome sequences for 14 unrelated wild mouse samples. I'd like to use GIBDLD to scan for selective signitures because it looks to me it's a very powerful method. I tried to run on chromosome 10 which has over 700,000 SNPs, I used -hiddenstates = 3 and -ibd = 3. Out of 105 pairs, I got identity coefficient = '-nan' for all SNPs for 51 pairs when I used unphased data, while I got '-nan' for 43 pairs when I used phased data (Inferred by Beagle). Those pairs are random and didn't invlove...