Fig. 4.

Disk space usage for the gPBWT versus sample count for GRCh38 chromosome 22. Points are sampled at powers of two up to 128, and intervals of 128 thereafter up to 1024. The trend line shown corresponds to the function $y(x)=3.16\times {10}^6\mathrm{bytes}·ln(x/\mathrm{samples})+5.12\times {10}^4\frac{\mathrm{bytes}}{\mathrm{sample}}·x+1.84\times {10}^8\mathrm{bytes}$.