Low-coverage BAMs share reads across individuals The 20 individuals in each BAM set share most of their simulated reads. In the first 200 kb of the Bottleneck Testing BAMs, 2,453 of 2,638 reads in pop_1.bam recur with the same start, CIGAR and strand in the other 19 BAMs (pop_1_1 and pop_10_1 start at position 90 with 126M). Because sequencing errors recur with the reads, 755 of 4,591 non-reference base calls appear at identical depth in the 20 individuals. At position 703, each individual has two G reads and one A read. Site frequency spectrum methods count these errors as variants. The other low-coverage sets (Bottleneck Final, Split Isolation Testing and Final) show the same pattern, with reads at the same positions. The simulator may have reused one random seed. IBD segment file is missing two relationship classes The Testing segment file (syn75202632) omits the 100 half_sibling and 100 half_2nd_cousin answer pairs and their 400 individuals. The other classes are complete, and the file covers 989 of the 1,189 answer pairs. Because its IBD distribution resembles the Testing file's and shows no extra cluster near 1,700 cM, where half siblings would fall, the Final file (syn75202631) may have the same gap. Travis, would you be able to look into the BAMs and the Final segment file, and let us know if either will be updated?